354
Mechanism of Gas Generation in Lithium Ion Batteries by Overdischarge

Wednesday, October 14, 2015: 11:00
Remington A (Hyatt Regency)
M. Hashimoto (Smart Energy Research Laboratories, NEC Corp.), M. Yamashiro, T. Ichihashi (Tokyo University of Science), A. Toda (Smart Energy Research Laboratories, NEC Corp.), T. Miyazaki (Smart Energy Research Laboratories, NEC Corp.), and S. Fujieda (Smart Energy Research Laboratories, NEC Corporation)
To ensure safety, lithium ion batteries (LIBs) are used with control circuits that restrict their operation voltages within an appropriate range. However, because of the capacity variation among LIB cells and the potential failure of battery control circuits, the behavior of LIBs at voltages higher or lower than the safe range should be fully clarified. Previous studies have focused primarily on over-charge and less on over-discharge (ODC), but the risk of ODC seems to be growing due to the wide spread and long-term use of LIBs. ODC has been reported to generate gas, but the gas species detected have varied. It has also been reported that ODC causes Cu to dissolve from the anode current collector, and that deep ODC would destroy SEI films.

We examined the structural change of the SEI film by ODC as well as the gas species generated to elucidate how ODC generates gas. Furthermore, we performed isotopic labeling of the electrolyte to find the origin of the gas.

We prepared LiMn2O4/carbon-based LIBs with aluminum-laminate packaging by using LiPF6 and an electrolyte consisting of EC and DEC. A fresh cell and a cycled cell with a capacity that had degraded to 40% were discharged to 0.5 V. The amount of ODC-generated gas was estimated from the cell-volume change. The gas volume was 2.1 cc in the fresh cell and 33.3 cc in the degraded cell, indicating that degraded cells suffer from a higher amount of gas evolution. The generated gas was analyzed by GC-TCD (carrier gas: Ar) and QMS. The generated gas was dominantly H2 (>75%), including a small amount of CxHy and COz molecules, which was in contrast to previous studies. Moreover, QMS detected a trace of CuCO, which might be produced by a reaction between the dissolved Cu atoms and the electrolyte. The difference in the major gas species may be due to the sample materials but could also be due to the gas analysis conditions such as the carrier gas species and mass scan ranges. 

The thickness of the SEI films in the anode was evaluated by AES with Ar+ sputtering. The SEI thickness, which we define as the depth at which the carbon proportion reaches 50%, was about 40 nm before ODC and 25 nm after ODC. Additionally, XPS measurements on the anodes revealed that ODC reduced the Li-related SEI components. These results indicate that ODC causes SEI to decompose. We also confirmed with cross-sectional SEM-EDX that ODC roughened the Cu anode current collector, thus causing Cu atoms to diffuse into the carbon active materials.

To investigate the origin of the hydrogen involved in the generated gas, we prepared two types of cells with deuterated EC (series A) and standard hydrogenated EC (series B). We cycled them at 45oC until their capacity faded to 50%. Since EC is known to decompose more easily than DEC, the series-A samples were expected to have SEI films containing D atoms after the cycling, which we confirmed by cross-sectional TOF-SIMS.  We also prepared a third type of cells (series C) by replacing the electrolyte of the cycled series-B cells with that containing deuterated EC. The three types of cells were discharged to 0.5 V, and the evolved gas was analyzed by QMS. In the series-A samples, a small amount of HD and D2 molecules were detected, but the majority consisted of H2 molecules. In the series-B samples, only H2 molecules were detected as expected. The molecules generated in the series-C samples were primarily H2. These results indicate that the origin of the generated H2 is not the SEI components derived from the electrolyte (not from EC, at least), and that the H2-gas evolution is not a result of direct decomposition of the electrolyte. The origin of the H2 is presumably hydrogen that is derived from the cathode and incorporated into the SEI on the anode.

In summary, we found that ODC destroys SEI films on the anode and generates H2 gas. The H2 does not seem to be derived from the electrolyte (EC); rather, H-impurities in the cathode active material are likely to be the H2 source. The volume of the ODC-induced gas increases with cycles. This tendency is attributable to a thicker SEI with H atoms and a higher anode potential upon ODC of the degraded cells. Suppressing the residual H-concentration in the cathode and SEI growth seems to be an effective way to eliminate the gas generation upon ODC.