Though the solid electrolyte interface (SEI) is critical to the long term stability of Li-ion batteries, its nature remains elusive. X-ray and electron-based probes are relatively insensitive to the low atomic number constituents of the SEI. Neutrons, however, are scattered by the nuclei of atoms rather than by the electrons, and are thus a relatively sensitive probe of structures containing light elements such as hydrogen and lithium. In situ neutron reflectometry (NR) measurements of SEI formation on a tungsten thin film model electrode will be presented. Building upon a previous study, tungsten was selected for its scattering length density (SLD), which had an optimal contrast with the solution for highlighting the growth of a low SLD layer at the interface.¹

NR of the bare electrode was first measured in an inert atmosphere, after which the electrode was assembled into a cell with a 1 mol/L LiPF₆ solution in ethylene carbonate + diethyl-d₁₀ carbonate (1:1 vol:vol). NR data were then measured at open circuit (about +2.65 V vs Li/Li⁺), and at a sequence of three fixed potentials: +0.25 V, +2.65 V, and again at +0.25 V vs Li/Li^+­­after cycling the electrode potential 10 times between +2.65 V and +0.25 V.

The NR data were fit to yield a SLD depth profile (in the direction normal to the interface) from which layer thicknesses are determined with sub-nanometer accuracy and elemental/isotopic composition can be inferred. A thin surface layer, presumably oxide, was present on the bare electrode. This was slightly modified by contact with the solution at open circuit, possibly indicating a reaction with the solution. When the electrode was then held at +0.25 V, the NR data were best fit by a model with two surface layers: an inner layer of low SLD, and an outer layer with an SLD close to that of the solution. The low SLD of the inner layer suggests a high content of Li and/or ¹H nuclei, as might be consistent with LiOH or underpotential deposited lithium. The higher SLD of the outer layer could be consistent with species richer in ²H, C, O, F, or P, such as might be derived from LiPF₆ or diethyl-d₁₀carbonate.

Upon increasing the potential to +2.65 V, the SLD of the inner surface layer increased markedly, indicating loss of Li and/or ¹H containing species (and possibly with solution filling the void spaces left by their loss). After potential cycling and returning to +0.25 V, the NR could be fit with a single surface layer with a SLD that decreased but not as far as the first hold at +0.25V. The thickness of this layer was about the same as the two layers present during the first potential hold at +0.25 V. The change from a two to a one surface layer model may be indicative of greater roughness between two surface layers, preventing NR from resolving them.

These data confirm the earlier measurements of SEI thickness and that the SEI is not immutable once it has been deposited. They also demonstrate the increased sensitivity of W electrodes for elucidating the structure of the SEI. In future work, we plan to apply isotopic substitution to further define the composition of the SEI and to study the effects of electrolyte additives in this model system.

1. Owejan, J. E.; Owejan, J. P.; DeCaluwe, S. C.; Dura, J. A., Chem. Mater. 2012, 24 (11), 2133-2140.