Compressed Sensing for X-ray Elemental Analysis
Compressed sensing (CS) is a relatively recent advancement originating from the mathematical and signal processing communities that allows for the acquisition and reconstruction of signals from vastly reduced amounts of collected data. Already, CS has given rise to breakthroughs in photography (like the Single pixel camera) and in medical imaging (drastically reducing the time needed for an MRI).
More recently, these techniques have been used in the electron microscopy community to enable three-dimensional TEM reconstructions from vastly reduced numbers of images,1 reducing not only the amount of time required for one of these experiments, but also the total electron dose that the sample receives (which for some materials, can be very important in order to prevent damage).
These methods have also been applied in scanning TEM imaging,2 3 as well as in the scanning electron microscope.4 5 These works have shown promising results in reducing both the time and electron dose needed for electron imaging.
Our work focuses on an expansion of these strategies into the analytical realm. Composition maps acquired using electron x-ray dispersive spectroscopy (EDS) are often extremely undersampled in fast-mapping conditions, and appear to be a ripe area for application of compressed sensing techniques. Our early results indicate that this is a very promising technique, and should be able to vastly improve the speed with which samples can be compositionally mapped in the SEM or TEM using EDS.
- Saghi et al., Nano Letters, 11, p. 4666 (2011). ^
- Stevens et al., Microscopy, 63, p. 41 (2014). ^
- Kovarik et al., Applied Physics Letters, 109, p. 164102 (2016). ^
- Anderson et al., Proc. SPIE 8657, Computational Imaging XI, p. 86570C (2013). ^
- Hujsak et al., Microscopy and Microanalysis, 22, p. 778 (2016). ^
Joshua Taillon
Materials Data Scientist
A materials research scientist at NIST interested in scientific data curation, AI for materials research, and baking bread.