Field emission EPMA is ideal for characterising optical fibres, which tend to be small and have complex cores with zonation on the micron and sub-micron scale.
Whilst optical fibres can be cleaved, to mount the sample flat for EPMA analysis the method we used was to embed them in expoxy and polish them. This is difficult optical fibres are thinner than a hair! The trick is to mount them perpendicular so that the cross section is circular. The other difficulty is adhession to the expoxy so that they dont move during polishing and create a poor contact; being non-conductive it is critical that the chage can be removed via a conductive coat, requiring a good contact between the optical fibre and the expoxy. Best results were achieved when the outer plastic coating was removed.
Optical fibres typical contain a range of dopants, and to achieve high resolution low voltages or multiple votlages may be required. In the case of Ge we have the choice of the Kα x-ray line or the Mα x-ray line. When considering the lateral resolution the spot size of the beam has to be considered, which increases at lower voltages and high beam currents, as described in the linked open access article by Pinard et al. (2014).
To obtain information regarding the composition of the core, x-ray maps work well ensuring that the centre of the core is located and avoiding convolution from offset analysis points. The x-ray maps can be quantified and the information extracted using line profiles, or radial extract. Radial extract - where the data is extracted as a series of expanding circles - is ideal for optical fibres because they are circular and maximises the counts utilized. A great plugin was developed for imagej - see here.
The graphs below compare radial and line extract for Ge. The radial extract improves precision, although errors are largest in the centre of the circle - where only several pixels are averages compared to the outside of the circle where the whole circumference is averaged.