X-ray microanalysis plays a key role in the development, production and failure analysis of metals and advanced metallic materials due to its ability to examine samples down to the nanometer scale. This valuable technique provides elemental and chemical information about the sample to supplement the morphological information provided by the electron microscope.
In part 1 of this post we discussed why elemental analysis combined with multivariate statistical analysis enables more confident conclusions than with elemental mapping alone. This post will describe a specific example of a more conclusive inference that can be made from a sample by conducting spectrum-focused analysis on a pixel-by-pixel basis.
A production turbine blade was sectioned and polished for energy dispersive spectrometry (EDS) analyses in the scanning electron microscope. The electron image showed sub-micron particles in a band parallel but distant to the surface of the blade. The challenge was to fully characterize the particles in the sample at low SEM beam energies. A spectral imaging data set was collected at 5 kV, 256 × 198 pixels. The nominal composition of the alloy was Ni-20Cr-12Al-4Mo.
Elemental maps of the known elements were displayed. The bands on the right and left of the particles had varying amounts of the primary elements (Ni, Cr, Al) with the region between the particles having a third composition. The particles in the central band were found to have a high concentration of Mo. The results appear quite conclusive; however, advance processing techniques including deconvolution and multivariate statistical analysis indicated the presence of another particle phase.
Using multivariate statistical analysis software, an analysis was performed on the same data and provided map-spectrum pairs for each of the chemically unique phases in the sample. However, instead of the expected four phases as found by the elemental map method (three bands of Ni-Cr-Al and Moenriched particles), another particle phase was presented (Figure 3 in the study). Close examination of the phase spectrum of these new particles finds that these particles also had an enrichment of Mo-L and also a higher enrichment of Nb-L (Figure 4). This Nb contribution was easy to miss in the original data set because of the heavy overlap of the Nb-L lines with the Mo-L lines, the very low composition of the Nb in the alloy, and the very few number of pixels contained within the Nbenriched phase.
When the results from the software analysis found that there is an enrichment of Nb in the particles, that element was added to the elemental map list and quantitative map results were produced (Figure 5). The deconvolution routines were able to separate the contributions of the Nb-L and the Mo-L X-rays in the overlapped peak and provide elemental maps that showed the true spatial distribution of the elements. The Nb-L quantitative map reinforced the multivariate statistical analysis software phase map results that not all of the particles are exclusively Mo but some of the particles are enriched with Nb. Careful measurement of these particles indicates that they are between 0.3 and 0.5 μm in width.
To see images and spectra obtained during the study, download Trends in X-ray Microanalysis: Enabling Rapid Discovery One Pixel at a Time.
For another example, read Rapid Evaluation of Smelting Copper Compounds Using Automated Multivariate Statistical Phase Analysis of EDS Spectral Imaging Data, which demonstrates that by using multivariate statistical analysis software, you can easily and rapidly determine the morphology, distribution and chemistry of each distinct compound generated in the initial process of copper lead smelting. This led to adjustment of the refining process and a cost savings to the smelting process.