Fluoroethylene carbonate (FEC) and vinylene carbonate (VC) are widely used as electrolyte additives in lithium ion batteries. Here we analyze the solid electrolyte interphase (SEI) formed on binder-free silicon nanowire (SiNW) electrodes in pure FEC or VC electrolytes containing 1 M LiPF by solid-state NMR with and without dynamic nuclear polarization (DNP) enhancement. We find that the polymeric SEIs formed in pure FEC or VC electrolytes consist mainly of cross-linked poly(ethylene oxide) (PEO) and aliphatic chain functionalities along with additional carbonate and carboxylate species. The formation of branched fragments is further confirmed by C-C correlation NMR experiments. The presence of cross-linked PEO-type polymers in FEC and VC correlates with good capacity retention and high Coulombic efficiencies of the SiNWs. Using Si DNP NMR, we are able to probe the interfacial region between SEI and the Si surface for the first time with NMR spectroscopy. Organosiloxanes form upon cycling, confirming that some of the organic SEI is covalently bonded to the Si surface. We suggest that both the polymeric structure of the SEI and the nature of its adhesion to the redox-active materials are important for electrochemical performance.

To elucidate the role of fluoroethylene carbonate (FEC) as an additive in the standard carbonate-based electrolyte for Li-ion batteries, the solid electrolyte interphase (SEI) formed during electrochemical cycling on silicon anodes was analyzed with a combination of solution and solid-state NMR techniques, including dynamic nuclear polarization. To facilitate characterization via 1D and 2D NMR, we synthesized C-enriched FEC, ultimately allowing a detailed structural assignment of the organic SEI. We find that the soluble poly(ethylene oxide)-like linear oligomeric electrolyte breakdown products that are observed after cycling in the standard ethylene carbonate-based electrolyte are suppressed in the presence of 10 vol% FEC additive. FEC is first defluorinated to form soluble vinylene carbonate and vinoxyl species, which react to form both soluble and insoluble branched ethylene-oxide-based polymers. No evidence for branched polymers is observed in the absence of FEC.

Over the last two decades solid state Nuclear Magnetic Resonance has witnessed a breakthrough in increasing the nuclear polarization, and thus experimental sensitivity, with the advent of Magic Angle Spinning Dynamic Nuclear Polarization (MAS-DNP). To enhance the nuclear polarization of protons, exogenous nitroxide biradicals such as TOTAPOL or AMUPOL are routinely used. Their efficiency is usually assessed as the ratio between the NMR signal intensity in the presence and the absence of microwave irradiation εon/off. While TOTAPOL delivers an enhancement εon/off of about 60 on a model sample, the more recent AMUPOL is more efficient: >200 at 100 K. Such a comparison is valid as long as the signal measured in the absence of microwaves is merely the Boltzmann polarization and is not affected by the spinning of the sample. However, recent MAS-DNP studies at 25 K by Thurber and Tycko (2014) have demonstrated that the presence of nitroxide biradicals combined with sample spinning can lead to a depolarized nuclear state, below the Boltzmann polarization. In this work we demonstrate that TOTAPOL and AMUPOL both lead to observable depolarization at ≈110 K, and that the magnitude of this depolarization is radical dependent. Compared to the static sample, TOTAPOL and AMUPOL lead, respectively, to nuclear polarization losses of up to 20% and 60% at a 10 kHz MAS frequency, while Trityl OX63 does not depolarize at all. This experimental work is analyzed using a theoretical model that explains how the depolarization process works under MAS and gives new insights into the DNP mechanism and into the spin parameters, which are relevant for the efficiency of a biradical. In light of these results, the outstanding performance of AMUPOL must be revised and we propose a new method to assess the polarization gain for future radicals.