Recently, it was shown that H₂O contamination in lithium-ion batteries can cause H₂ gassing during the formation of the solid-electrolyte interphase (SEI) on graphite as well as substantial concentrations of CO₂ in the battery cell.¹ The latter was attributed to the hydrolysis of ethylene carbonate (EC) in the electrolyte by OH^- formed through H₂O reduction at the anode. The H₂O- and OH^--driven EC hydrolysis rates were measured at temperatures between 10 and 80°C by adding discrete amounts of H₂O or TBAOH•30H₂O to the electrolyte and monitoring the associated CO₂ evolution by On-line Electrochemical Mass Spectrometry (OEMS).²

Apart from the intrusion of H₂O into the battery cell as a trace contaminant, it was recently reported that H₂O could be generated from the reaction of carbonate electrolyte with reactive oxygen released from NMC cathodes at high voltage.³ Thus, the often detected formation of POF₃ at high-voltage⁴ could be explained by the reaction of PF₅ with in-situ generated H₂O to yield POF₃ and HF. However, a competing reaction would be the reduction of the generated H₂O at the anode (crosstalk⁵) releasing H₂ and OH^-, and forming CO₂ via the hydrolysis of EC.

In this study, we will deconvolute these reactions. First, we present an accurate quantification of POF₃ by determining its calibration factor in OEMS using a high-temperature cell and employing the thermal decomposition of LiPF₆ to POF₃ and PF₅, as previously evidenced by TGA-FTIR.⁶ We further use accelerated tests to determine the thermal reactivity of both, H₂O and OH^-, with EC in a LiPF₆ containing electrolyte, yielding CO₂ and POF₃. Next, we correlate our accelerated test with the POF₃ and CO₂ evolution formed by electrochemical side reactions during the SEI formation on graphite in presence and absence of H₂O and OH^-. For this, we employ our recently presented two-compartment cell in which anode and cathode compartments are separated by a Li⁺-ion conducting solid electrolyte (Ohara glass) laminated with aluminum and polypropylene foil, so that the gas evolution from side reactions at the graphite electrode can be studied without cross-diffusion of reaction products to the counter-electrode, thereby enabling a more detailed analysis of the reaction pathways.⁵ This is a major advance over conventional cells, where the gases come from both electrodes, and thus do not allow a deconvolution of the simultaneously occurring reactions from anode and cathode. The study is concluded by discussing the impact of the contaminants, H₂O and OH^-, on the cycle life of NMC//graphite full-cells at 45°C where both side reactions, the hydrolysis of EC and the decomposition of LiPF₆, can play a role.

Figure 1 shows an exemplary OEMS measurement with a 1.5 M LiPF₆ EC electrolyte containing either H₂O (blue) or OH^- contamination (green) in an accelerated test to determine the gas evolution from side reactions with these contaminants at various temperatures.

References

1. R. Bernhard et al., J. Electrochem. Soc., 162, A1984 (2015).
2. M. Metzger et al., J. Electrochem. Soc., 163, A1219 (2016).
3. R. Jung et al., 231^st ECS Meeting, #39, New Orleans, USA (2017).
4. M. Metzger et al., J. Electrochem. Soc., 162, A1227 (2015).
5. M. Metzger et al., J. Electrochem. Soc., 163, A798 (2016).
6. H. Yang et al., J. Power Sources, 161, 573 (2006).

Acknowledgement

The authors gratefully acknowledge BASF SE for financial support of this research through the framework of its Scientific Network on Electrochemistry and Batteries.

Figure 1. Temperature set point (black) and cell temperature (red) vs. time during OEMS measurements (upper panel), and respective CO₂ and POF₃ evolution (lower panel) for 240 μl of 1.5 M LiPF₆ in EC with H₂O (blue) or OH^- contamination (green). The gray data shows a reference measurement with only the base electrolyte (<20 ppm H₂O).