The Na-O₂ battery is a promising energy storage device due to its high theoretical energy density and low cost of sodium metal, however this technology remains in the early stages of development, due to the challenges related to reversibility, and electrolyte stability.^[1]  Here we demonstrate the utility of ²³Na nuclear magnetic resonance (NMR) as a diagnostic tool for screening sodium–oxygen (Na-O₂) discharge products. When an ether-based electrolyte is employed, either one or both of the desirable products; sodium superoxide (NaO₂) and sodium peroxide (Na₂O_2­) is electrochemically formed.^[2-5] In addition to the anticipated electrochemistry, electrolyte breakdown also occurs during the operation of the cell, where sodium carbonate (Na₂CO₃) is a main electrolyte breakdown product.^[1]Currently the underlying battery chemistry is still unclear but can be revealed through the careful characterization of electrochemically-cycled electrodes, using advanced strategies including solid-state NMR.

NaO₂, Na₂O₂ and Na₂CO₃ are readily distinguishable in the ²³Na NMR spectrum, as shown in the experimental data in Figure 1, and supported by our quantum chemical calculations of the quadrupole parameters for both ¹⁷O and ²³Na.  In a mixed sample, where the potential reaction products are ground together, the presence of paramagnetic NaO₂ is observable based on its well-resolved chemical shift and rapid relaxation. This is contrasted by the spectral overlap and slow relaxation times of the diamagnetic Na₂O₂ and Na₂CO₃ species, which have both unique, and overlapping resonances in their ²³Na MAS NMR spectra. Nevertheless, the two species can be quantified using 2D multiple quantum spectroscopy applied under magic angle spinning (MQ-MAS) at 20-40kHz.  We have utilized this NMR strategy to develop calibration curves for the relative intensities of the peaks in 2D spectra, as a function of the mass of each possible reaction product.   We have compared this data with the 2D MQMAS spectra acquired for electrodes extracted from sodium-oxygen cells, as shown in Figure 2.  Three regions of peaks are clearly visible, with the broad, lower frequency peak assigned to sodium carbonate, and the narrow, highest frequency peak assigned to sodium peroxide.  The challenging region to interpret is the overlapping middle region, where both sodium peroxide and sodium carbonate have resonances, as evident also in Figure 1.  The 2D NMR strategy used here allows for quantitative extraction of the individual quadrupolar lineshapes, which compare very well to the calculated lineshapes for these constitutents.  Thus, the calibration data can be used to systematically compare electrochemical reaction products for the sodium-oxygen electrochemistry.

This investigation begins with a comparison of electrochemically-cycled electrodes where the electrolyte was diethylene glycol diethyl ether. With ²³Na NMR we can determine which reaction product is electrochemically formed in Na-O₂cells as a function of both electrolyte composition and cycling conditions.

[1]        Q. Sun, Y. Yang, Z.-W. Fu, Electrochemistry Communications 2012, 16, 22.

[2]        P. Hartmann, M. Heinemann, C. L. Bender, K. Graf, R.-P. Baumann, P. Adelhelm, C. Heiliger, J. r. Janek, The Journal of Physical Chemistry C 2015, 119, 22778.

[3]        W. Liu, Q. Sun, Y. Yang, J.-Y. Xie, Z.-W. Fu, Chemical Communications 2013, 49, 1951.

[4]        C. L. Bender, W. Bartuli, M. G. Schwab, P. Adelhelm, J. Janek, Energy Technology 2015, 3, 242.

[5]        H. Yadegari, Y. Li, M. N. Banis, X. Li, B. Wang, Q. Sun, R. Li, T.-K. Sham, X. Cui, X. Sun, Energy & Environmental Science 2014, 7, 3747.