Entropy Hysteresis during Lithiation/Delithiation of NCA/Gr-Si Battery Subjected to Accelerated Calendar Ageing and Cycle Ageing

Abstract

The literature surrounding entropy changes accompanying degradation is scarce and limited to solely graphite anode cells. Meanwhile, graphite-silicon blends become frequent in commercial applications due to their considerable capacity advantage. The lithiation/delithiation results in volume changes of the silicon particle, which has been reported to cause an increased hysteresis [1] of the open circuit potential (OCP). Our hypothesis is that entropy reflects certain morphological changes occurring within the electrode and consequently, entropy hysteresis is also higher for electrodes containing silicon. We further postulate that entropy hysteresis increases with cycle age. If the hypothesis is correct, entropy measurement will offer a unique insight into battery degradation among commonly used differential voltage analysis (DVA) and incremental capacity analysis (ICA).To test our hypothesis, we adapted an accelerated entropy measurement method proposed by Osswald et al. [2] on high-energy NCA/Gr-Si cylindrical cells, with ~10 wt % Si and ~90 wt % Gr anode composition. The cells were divided into two groups; the first group was stored at an elevated temperature to act as an example of accelerated calendar ageing, while the second group experienced cycle ageing. Subsequently, we performed DVA and ICA to provide a direct comparison with the entropy results and checked for correlation.In accordance with the hypothesis, the entropy behaved similarly to the OCP. Entropy hysteresis remained stable for calendar aged cells (Fig. 1 a) but increased considerably for cycled cells (Fig. 1 b). Silicon volume expansion and its 'breathing' effect [3,4] caused charge entropy to increase with cycle age. Graphite particles experienced breaking and cracking, which prompted a decrease in discharge entropy during cycling. These combined effects led to the observed rise in entropy hysteresis over time.A direct comparison of entropy profiling with DVA revealed alike characteristics. Based on abrupt energy level changes accompanying phase transitions, entropy profiling was successfully used to track ageing markers, aiding recognition of a loss of active material on positive (LAMPE) and negative (LAMNE) electrodes as well as loss of lithium inventory (LLI). Both DVA and entropy profiling revealed that LLI was the main degradation mode for the calendar aged cell, while LAMNE combined with LLI for the cycled cell. Plotting entropy against voltage allowed for additional observations. Horizontal shift towards higher voltages occurred due to the rise in internal resistance but also LLI. While some authors [5] successfully obtained information about LAMPE and LAMNE from ICA, and an analogy can be performed for entropy profiling, it is difficult to draw definitive conclusions from these results.The fact that entropy profiling reflects microscopic changes occurring within electrodes, and considers also ageing markers, makes it a unique, non-invasive tool among ICA and DVA. However, its application is not straightforward and needs further validation. A possible avenue to be explored is the theoretical simulation of pristine and aged entropy profiles to cross-validate with our experimental data.