Enhanced resolution of soft-materials spectroscopic imaging in the scanning transmission electron microscope

The quantitative analysis of soft-materials morphology at nano lengths is an important scientific and technical challenge. Imaging based on spatially resolved Electron Energy-Loss Spectroscopy (EELS) enables both real-space morphological measurements and the quantitative determination of local composition without assuming a particular model as is done by scattering approaches. EELS imaging is being increasingly used in a variety of hard-materials applications. However, its application to soft materials, such as synthetic polymers and biological tissue, remains challenging because of the resolution limits imposed by the radiation sensitivity of most soft materials. This thesis explores the factors that affect the dose-limited resolution of soft materials, and it develops new approaches to improve this resolution. We show that the accuracy of compositional analysis can be compromised in order to enhance the resolution, and we successfully apply this approach to a semi-quantitative analysis of alkane-based coatings on nanosized poly(amine) nanoparticles. More generally, however, one would like to preserve compositional accuracy while using the higher electron doses required to achieve high resolution. To this end, we have discovered that the effects of radiation-induced evolution of hydrogen---a damage mechanism known to be a significant limitation in EELS studies of frozen-hydrated soft materials---can be completely avoided if very thin TEM sections are studied. We illustrate the thickness dependence of hydrogen evolution in solvated Nafion, a perflourinated ionomer, and in hydrated porcine skin. Then, working with thin sections of frozen-hydrated skin, we develop and apply a method to extract from an experimental spectrum dataset a reference spectrum that accurately represents the hydrated skin's protein component under conditions where the protein has suffered significant radiation damage. Using such an extracted reference spectrum virtually eliminates the error associated with a multiple least squares compositional analysis. We achieve 10 nm resolution in fully quantitative and accurate maps of the water distribution. This is almost an order of magnitude better resolution than previously achieved from tissue using separately collected reference spectra, and it opens the opportunity for new advances in measuring the nanoscale spatial distribution of water and its role in a variety of water-mediated phenomena, in skin, other biological tissue, and a range of hydrated synthetic polymers.