Multivariate statistical analysis as a tool for the segmentation of 3D spectral data

Acquisition of three-dimensional (3D) spectral data is nowadays common using many different microanalytical techniques. In order to proceed to the 3D reconstruction, data processing is necessary not only to deal with noisy acquisitions but also to segment the data in term of chemical composition. In this article, we demonstrate the value of multivariate statistical analysis (MSA) methods for this purpose, allowing fast and reliable results. Using scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDX) coupled with a focused ion beam (FIB), a stack of spectrum images have been acquired on a sample produced by laser welding of a nickel–titanium wire and a stainless steel wire presenting a complex microstructure. These data have been analyzed using principal component analysis (PCA) and factor rotations. PCA allows to significantly improve the overall quality of the data, but produces abstract components. Here it is shown that rotated components can be used without prior knowledge of the sample to help the interpretation of the data, obtaining quickly qualitative mappings representative of elements or compounds found in the material. Such abundance maps can then be used to plot scatter diagrams and interactively identify the different domains in presence by defining clusters of voxels having similar compositions. Identified voxels are advantageously overlaid on secondary electron (SE) images with higher resolution in order to refine the segmentation. The 3D reconstruction can then be performed using available commercial softwares on the basis of the provided segmentation. To asses the quality of the segmentation, the results have been compared to an EDX quantification performed on the same data.

Nd:YAG laser joining between stainless steel and nickel-titanium shape memory alloys

Aïcha Hessler-Wyser, Michel Rappaz, Jonas Vannod

Nickel-Titanium (NiTi) shape memory alloys are often used in medical component devices, for instance as guide wires for neurological surgery applications. The manufacture of such devices becomes more and more challenging, especially considering the need to join them with other metals. Laser welding is a promising technique to realize and to guaranty the mechanical stability of dissimilar metal welds, although inherent differences in chemical compositions, absorption, physical and thermo-mechanical properties can lead to severe problems, in particular fracture of the weld due to the possible formation of brittle intermediate phases. Laser welds of NiTi - stainless steel (SS) pieces have been made with a Nd:YAG laser and the weld microstructure have been studied by Scanning Electron Microscopy (SEM). The phases and defects in these welds have been compared with those observed in autogenous NiTi welds and SS welds. They have been put into relation with the Fe-Ni-Ti ternary phase diagram. In addition, Differential Thermal Analyses (DTA) of NiTi-SS alloys and NiTi-SS diffusion couple experiments have been performed in order to gain a better understanding of the phases and reactions occurring during laser welding. This diffusive couple experiments have been analyzed by Energy Dispersive X-ray Spectroscopy (EDX) It appeared that although experiments were performed under controlled atmosphere, oxide layers have restrained the chemical diffusion of concerned elements. Further diffusion couple experiments will be realized with stainless steel welded caps to avoid oxygen contamination during heating.

2008

Three Dimensional Microanalysis by Energy Dispersive Spectrometry

Pierre Burdet

This thesis is focused on a combined microscopy technique: energy dispersive spectrometry (EDS) is extended to a three dimensional (3D) microanalysis using a so-called "dual-beam microscope": a scanning electron microscope (SEM) equipped with a focused ion beam (FIB). In the sequential acquisition, the surface freshly milled by the FIB is characterised by SEM imaging and EDS mapping. A 3D elemental picture of the specimen is obtained this way. This technique suffers from the same limitations than the 2D EDS mapping, the major one being linked to the volume of X-ray emission that is large due to the required high accelerating voltage. Other limitations of 3D EDS microanalysis are more specific to FIB/SEM technique, such as the low acquisition time per spectrum due to the large number of spectra required in an acquisition. The goal of this thesis was to develop post-processing solutions to overcome the limitations of 3D EDS microanalysis. Three solutions have been developed. As the acquired data are composed of a high number of noisy spectra, multivariate statistic methods are appropriate. Such a technique is adapted to 3D EDS data and provides smoother spectra improving the quantification afterwards. When analysing a feature that is smaller than the volume of X-ray emission, the quantified composition is inaccurate as part of the X-rays are emitted from the feature’s surrounding. To take into account the influence of the neighbouring voxels, an enhanced quantification technique is developed. It is based on a recursive approach adapting an existing complex quantification. Another complementary approach is developed to resolve features too fine for EDS mapping: the segmentation technique is improved by using the higher spatial resolution of SEM images. A sample formed by laser welding of nickel-titanium (NiTi) and stainless-steel wires is characterised by 3D EDS microanalysis. The acquired data are used to demonstrate the gains and the limitations of the three developed processing techniques. With them, the noise-reduced spectra reveal further details of the fine microstructure. Their quantified composition is closer to the one predicted by the phase diagram. Furthermore, the segmented phases used for the 3D visualisation have a resolution close to the one of the SEM images. This visualisation allows a deeper comprehension of the formation of the phases and their morphologies during the implied solidification. This demonstrates the great potential of this technique to characterise samples with complex microstructure and complex composition.

EPFL2012

Three Dimensional Microanalysis by Energy Dispersive Spectrometry

Pierre Burdet

EPFL2012