Water mapping in hydrated soft materials

We present a method based on spatially resolved electron energy-loss spectroscopy in the cryo-STEM to map the spatial distribution of water in frozen-hydrated polymers. The spatial resolution is limited by the dose constraints imposed by radiation damage, and to stay within these constraints, the use of fine electron-probe sizes comes at the cost of reduced counts in the energy-loss spectra. Thus, at the resolution limit, the detection of isolated water-rich pixels or the identification of minor variations in water content across the specimen is complicated because one must distinguish significant fluctuations from noise. Here we develop a criterion with which to guide such a distinction. We characterize the intrinsic noise associated with spectral measurements under given illumination and acquisition conditions. We then use that noise in combination with scatter diagrams to threshold spectrum images and objectively identify statistically significant compositional fluctuations. We illustrate these ideas using a simulated spectrum dataset for a hypothetical blend of hydrophilic and hydrophobic homopolymers. We show that while a direct inspection of the water map may not allow any meaningful conclusions to be drawn, after applying the thresholding approach we can clearly identify the regions of the specimen that are rich in water. We also experimentally study a model blend system comprised of hydrophilic poly(vinyl pyrrolidone) (PVP) dispersed in a hydrophobic matrix of poly(styrene) (PS). By MLS fitting using damaged and undamaged PVP reference spectra, we determine that the critical dose characteristic of dry PVP is ∼8000 e/nm2 using 200 keV incident electrons. Irradiating frozen-hydrated PVP gives rise to noticeable hydrogen evolution at doses of ∼1500 e/nm2. To stay within this constraint we use doses of 400 e/nm2 and a pixel spacing in the spectrum imaging of 100 nm. In order to quantitatively map the water, PVP, and PS compositions, we measure their total inelastic scattering cross-sections. Direct inspection of the composition maps reveals the presence of large water-rich domains of the order of ∼ 1 μm and the scatter-diagram thresholding approach identifies small water-rich domains one pixel in size.