Guy, M.L., et al., Chemisorption of Water on the Surface of Silicon Microparticles Measured by DNP-Enhanced NMR. The Journal of Physical Chemistry C, 2017.
We use dynamic nuclear polarization (DNP) enhanced nuclear magnetic resonance (NMR) at liquid helium temperatures to directly detect hydrogen attached to the surface of silicon microparticles. The proton NMR spectrum from a dry sample of polycrystalline silicon powder (1-5 μm) shows a distinctively narrow Lorentzian-shaped resonance with a width of 6.2 kHz, indicative of a very sparse distribution of protons attached to the silicon surface. These protons are within a few atomic monolayers of the silicon surface. The high sensitivity NMR detection of surface protons from low surface area (0.26-1.3 m2/g) particles is enabled by an overall signal enhancement of 4150 over the room temperature NMR signal at the same field. When the particles were suspended in a solvent with 80% H2O and 20% D2O, the narrow peak was observed to grow in intensity over time, indicating growth of the sparse surface proton layer. However, when the particles were suspended in a solvent with 20% H2O and 80% D2O, the narrow bound-proton peak was observed to shrink due to exchange between the surface protons and the deuterium in solution. This decrease was accompanied by a concomitant growth in the intensity of the frozen solvent peak, as the relative proton concentration of the solvent increased. When the particles were suspended in the organic solvent hexane, the proton NMR spectra remained unchanged over time. These results are consistent with the known chemisorption of water on the silicon surface resulting in the formation of hydride and hydroxyl species. Low-temperature DNP NMR can thus be used as a non-destructive probe of surface corrosion for silicon in aqueous environments. This is important in the context of using silicon MEMS and bioMEMS devices in such environments, for silicon micro- and nano-particle MRI imaging agents, and the use of nanosilicon for splitting water in fuel cells.