With the development of more energetic batteries for large scale applications, safety of Li-ion systems is becoming of a greater concern. Abuse tests such as over(dis)charge, nail test, short-circuit, crush, overheating are generally used to evaluate safety of Li-ion cells and determine the risk of having unwanted reactions such as swelling, leakage, explosion, fumes, fire… When tested in abuse conditions, stress is extreme and it intensifies chemical reactions and/or physical damages of internal components[1]. Some of these reactions have also been detected in Li-ion cells after long use in normal operating. Actually, battery thermal runaway can occur in non-abusive conditions originating from overpressure or internal short-circuit and main adverse phenomena experienced in Li-ion batteries are Li plating, gas generation due to material degradation and separator failure[2],[3]. In this communication, first we propose to show post-mortem characterizations of these three phenomena: - In most of our post-mortem investigation on aged cells, Li plating is still detected on negative side. We demonstrate the presence of Li metal on several graphite electrodes after ageing mainly using NMR and XPS analyses. Such deposition results from resistance increase that can be due to gas bubbles, porosity closing in separator, temperature decreasing, surface roughening … Recently, we examined one cell after long storage periods at fully charged state and few cycles. We demonstrate that electrolyte decomposed creating some gas, which modifies locally current distribution resulting finally in circular deposits of Li metal. - Other post-mortem analyses revealed also the degradation of polymer separator in commercial NMC/Graphite cell after calendar ageing at 45°C. Separator degradation has been evidenced by IR spectroscopy, microscopy and physical measurements. It became less porous and less elastic and Li diffusion is impacted (fig. 1a). Thus, maximum cycling rate has to be limited to avoid Li deposition and further uncontrolled reaction due to dendrites growth (fig. 1b). - In post-mortem investigation, analytical methods are necessary to characterize internal components of batteries and detect degradation by-products even in low concentration. Gas generated in Li-ion batteries can also be sampled and its composition is determined by chromatography coupled with IR. Gaseous by-products mainly result from electrolyte degradation or active material structural change. The second part of the communication will be focused on the methods we use for early detection of anomaly: i) potential monitoring with a reference electrode, ii) strain gauges and ii) electrochemical noise measurement. For Li plating, potential of anode is monitored thanks to a reference electrode implemented in a full Li-ion cell. That allows determining if Li metal deposition is likely to occur. For early detection of gassing in batteries, deformation can be measured using strain gauges on the packaging of the cell. As soon as degradation occurs in the cell, gaseous products will actually accumulate and cell will deform or swell. This gauge method can be more sensitive than conventional external thermal sensors. When gas is generated, the resistance of electrode/electrolyte interface is also directly impacted. As presented below (fig. 1c), electrochemical measurements do not show any evolution at LFP/electrolyte interface when stored at high temperature, whereas electrolyte resistance at LTO/electrolyte interface shows some increase with repeated sudden drops. This profile is attributed to bubbles formation and release at the interface. To conclude, as Li plating or separator ageing can lead to internal short-circuit and material degradation can provoke gas release and overpressure in the cell, degradation of internal components has to be considered for safety concern even in non-abuse conditions. Actual studies will help to enhance long-term stability and improve safety of Li-ion batteries. [1] A Review on Lithium-Ion Batteries Safety Issues: Existing Problems and Possible Solutions; Jianwu Wen, Yan Yu, and Chunhua Chen; Mater. Express, Vol. 2, No. 3, 2012 [2] M. Broussely et al. / Journal of Power Sources 146 (2005) 90–96 [3] M. Fleischhammer et al. / Journal of Power Sources 274 (2015) 432e439 Figure 1