Open Access

The transport and storage of lithium-ion (Li-ion) batteries — damaged or in an undefined state — is a major safety concern for regulatory institutions, transportation companies, and manufacturers. Since (electro)chemical reactivity is exponentially temperature-dependent, cooling such batteries is an obvious measure for increasing their safety. The present study explores the effect of cryogenic freezing on the electrochemical and physical stability of Li-ion cells. For this purpose, three different types of cells were repeatedly exposed to liquid nitrogen (LN2). Before and after each cooling cycle, electrical and electrochemical measurements were conducted to assess the impact of the individual freezing steps. While the electrochemical behavior of the cells did not change significantly upon exposure to LN2 , it became apparent that a non-negligible number of cells suffered from physical changes (swelling) and functional failures. The latter defect was found to be caused by the current interrupt device of the cylindrical cells. This safety mechanism is triggered by the overpressure of expanding nitrogen which enters the cells at cryogenic temperatures. This study underlines that the widely accepted reversibility of LN2 -cooling on a material scale does not allow for a direct extrapolation toward the physical integrity of full cells. Since nitrogen enters the cell at cryogenic temperatures and expands upon rethermalization, it can cause an internal overpressure. This can, in turn, lead to mechanical damage to the cell. Consequently, a more appropriate temperature condition — less extreme than direct LN2 exposure — needs to be found

The present work proposes the implementation of non-singular basis functions into the algorithm for reconstructing the distribution of relaxation times (DRT) function of impedance data. These functions reflect the dispersed and asymmetrical nature of non-ideal capacitive–resistive processes. Inclusion is achieved by combining the singular Debye distribution basis with distributed relaxation functions, such as those derived from the analytical models of Cole–Cole and Havriliak–Negami. The shapes of the introduced basis functions are described by constant parameters, for which an empirical optimization approach is provided alongside. Using synthetic impedance data of non-ideal capacitive–resistive processes subjected to white noise, it is shown that the demand for regularization can be reduced significantly by using distributed bases. To underline the practical relevance of non-singular basis functions in DRT reconstruction, an experimental study comprising 100 sodium-ion and 80 lithium-ion ommercial cells is presented. In this context, it is shown that auxiliary information from the non-ideal nature of real-world electrochemical processes is outsourced into the basis and, hence, easily filtered out of the resulting DRT. This facilitates the separation of single processes without post-DRT curve fitting and thus improves the interpretation and classification of impedance data significantly.