Knowledge of reversible and irreversible heat generation processes occurring in lithium-ion cells is inevitable for a proper design of thermal management systems. The determination of reversible heat is often carried out by means of a potentiostatic measurement of entropy related effects, which are based on structural properties of the active materials. This method utilizes the assumed linear relationship between a change in the open-circuit potential (OCP) of a cell for a change in its temperature. In this work, the potentiostatic method is more closely investigated regarding this linear relationship and possible associated measurement inaccuracies of entropic coefficients.
Various lithium-ion cells with differing chemistries and cell designs are investigated in this study. All cells were rested for more than six months at different States of Charge (SOC) at room temperature in order to guarantee a most equalized state. The chosen generic temperature excitation was composed of a positive and a consequent negative temperature pulse, both starting at a base temperature of 25 °C. Profile I is formed of temperature pulses with varying pulse length whereas within Profile II, the pulse amplitude was varied. After each temperature change, new entropic coefficients were calculated based on the preceding data. Additionally, the thickness of selected cells was monitored via 2D laser scan during Profile I and Profile II.
The results obtained in this work clearly reveal a non-linear behaviour of the voltage to temperature relationship regardless of the cell design, SOC, cell chemistry or sign of the derived entropic coefficient. This non-linearity manifests in a certain hysteresis of the OCP depending on the temperature history of the cells (i.e. Profiles I and II). Preliminary results from the thickness experiment also reveal a hysteresis in the thermal expansion, which we assume is the reason for the non-linear behaviour of the voltage.
An improvident determination of the entropic coefficients might lead to a relative deviation of up to 95%, resulting in inacceptable errors in the design of thermal management systems. In order to limit the measurement inaccuracies we suggest to minimize the temperature pulse duration whilst still allowing for a thermal equalization of the cell. Furthermore, we suggest to minimize the amplitude of the temperature pulse as much as possible, whilst still allowing for an acceptable Signal to Noise Ratio (SNR) for the measured voltage response.