Characterization of the solid electrolyte interface (SEI) in Li-ion batteries presents a challenge. Removing an electrode from a cell for ex situ measurements on the SEI risks its reaction with the atmosphere and also loss of material during washing. In situ characterization of the SEI is thus of great value. Neutron reflectometry (NR) enables such measurements with relatively high sensitivity to low atomic number elements, which may constitute a significant fraction of the SEI. Fitting of layer models to NR data yields a scattering length density (SLD) profile in the surface normal direction, from which a composition-depth profile may be inferred based upon the elements present in the cell.

In this work, the interfacial structure of magnetron sputtered tungsten thin film electrodes in contact with a LiPF₆ solution in an ethylene carbonate/diethyl carbonate (EC/DEC) solvent was characterized with NR. Tungsten was selected as a model electrode for its relatively low SLD contrast with the silicon substrate (relative to earlier measurements with a copper electrode¹) and because it is not expected to alloy with lithium. NR was first measured for the bare electrode in an inert atmosphere, then at open circuit in contact with solution, and at +0.25 V, +2.65 V, and +0.25 V vs Li/Li⁺ after potential cycling.

The bare electrode data were best fit with a four-layer model, including silicon oxide, an interfacial silicon oxide-tungsten layer, tungsten, and tungsten oxide layers. X-ray reflectivity data also supported this model. NR data collected at open circuit (about +2.5 V to 2.6 V) indicated a similar structure for the buried layers, but the tungsten oxide surface layer was no longer visible. The oxide layer may still have been present but more difficult to detect as a discrete layer because its SLD was intermediate between that of tungsten and the solution and because it was relatively thin and its interface width was comparable to the layer thickness.

NR data collected when the electrode was polarized to and held at +0.25 V vs Li/Li⁺ were best fit a with a model containing two SEI layers: an inner low SLD layer and an outer layer with a SLD close to that of solution. The inner layer may be inferred to consist of Li₂O or LiOH, given its very low SLD (the SLDs of other candidate species were too high to account for this layer). The inner layer SLD did not differ greatly in isotopic labelling measurements in which both EC and DEC were deuterated and where only DEC was deuterated. This indicated the inner SEI layer did not incorporate significant quantities of hydrogen atoms from EC, since the SLD of LiOD is significantly greater than that of LiOH. Since deuterated DEC can be a source of deuterons but not of protons, this suggests the layer was primarily Li₂O, barring any adventitious sources of protons. This could have been produced at least in part by reduction of the native tungsten oxide to Li₂O and metallic tungsten. The SLD of the outer layer did depend upon solvent deuteration, which indicates that it had solution-filled porosity or had EC-derived constituents (or both).

When the electrode was polarized to +2.65 V, the inner low SLD surface layer was removed, although an outer SEI layer of SLD close to that of the solution was still present. The electrode potential was then cycled between +2.65 V and +0.25 V ten times and then held at +0.25 V. NR data measured at this potential indicated a low SLD layer formed at the surface again, though its SLD was not as low as in the first +0.25 V hold. The two SEI layers were less readily discerned as independent layers than in the first hold at +0.25 V, suggesting roughening of the SEI.

These data show that the SEI is not immutable once formed, confirm thicknesses measured in earlier NR measurements, and show that the lower contrast of the thin film tungsten electrode may improve sensitivity of NR measurements to surface layers. Surface oxides of electrode materials, although very thin, can play an instrumental role in the formation and composition of the SEI and may lead to significant variations with differing electrode materials. In future work, we plan to use this platform to characterize the effects of electrolyte additives on SEI structure.

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