Comprehensive Calorimetry and Modeling of the Thermally-Induced Failure of a Lithium Ion Battery

Abstract

A lithium ion battery (LIB) subjected to external heat may fail irreversibly. Manifestation of this failure include venting of potentially combustible gases and aerosols followed by a rapid self-heating accompanied by ejection of the battery materials. Quantification and simulation of the dynamics and energetics of this process are important to ensure LIBs’ safety. Here we report on development of a new experimental technique for measuring the energetics of the thermally-induced failure of LIBs as well as a new thermo-kinetic model to predict battery failure behaviors. The newly developed experimental technique, Copper Slug Battery Calorimetry (CSBC), was employed to investigate a widely utilized form factor of LIB (i.e. 18650) with 3 different battery chemistries: lithium cobalt oxide (T-Energy ICR18650, LCO), lithium nickel manganese cobalt oxide (Panasonic CGR18650CG, NMC) and lithium iron phosphate (K2 18650E, LFP), at various states of charge (SOCs). This technique can yield time resolved data on the rate of heat production inside the failing battery. The heat capacity of these LIBs was evaluated to be 1.1±0.1 J g-1 K-1 for all three cathode types. It was shown that the total heat generated inside the batteries increases with increasing amount of electrical energy stored. The maximum total internal heat generated by fully-charged LIBs was found to be 37.3±3.3, 34.0±1.8 and 13.7±0.4 kJ/cell for LCO, NMC and LFP LIBs, respectively. Additionally, experiments were carried out in which the CSBC technique was combined with cone calorimetry to measure the heat produced in flaming combustion of vented battery materials. The released combustion heat did not show a clear dependence on the stored electrical energy; this heat varied between 35 and 63 kJ/cell for LCO LIBs, 27 and 81 kJ/cell for NMC LIBs, and 36 and 50 kJ/cell for LFP LIBs. Beyond the experimental work, detailed modeling of heat transfer in the CSBC experiments was carried out, by utilizing COMSOL Multiphysics software, to evaluate thermal conductivities of the LIBs and demonstrate the satisfactory accuracy of CSBC experimental analysis in the determination of the battery failure energetics for all examined battery types.

Moreover, it is presented in this study a general methodology to develop a thermo-kinetic model of thermally-induced failure of lithium ion batteries (LIBs), using COMSOL and experimental data collected by CSBC. This methodology is demonstrated specifically on LCO LIBs (T-Energy ICR18650), but it can be easily extended to other battery types. The model was parameterized based on Arrhenius’ Law and via an iterative inverse modeling analysis of CSBC test results using COMSOL. These model parameters are dependent on the cells’ states of charge (SOCs) and they can effectively represent the tested cells’ heat production energetics during failure. The fully-parameterized thermo-kinetic model was then validated against CSBC tests that were not utilized in the model parameterization: CSBC tests on 100% SOC LIB cell with non-standard heating rates ranging from 40 W to 100 W; and CSBC tests on 75% SOC LIB cell with a standard heating rate of 20 W. The agreements between the experimentally measured and the simulated copper slug temperature histories in these tests were found within in 5% on average. Last but not least, this model was applied to predict the thermally-induced failure of LIB cells in a more complex scenario – cascading LIB failure of 6 LIB cells in a billiard battery pack. The simulated onset time of thermal runaway of each LIB cell in the battery pack were found of excellent agreements with experimental observations.