Characterisation and prediction of state of health of rechargeable batteries

Abstract

Electrochemical impedance spectroscopy (EIS) is used extensively in battery research, but usually at frequencies that do not reflect real-world usage patterns. This has resulted in equivalent circuit models (ECMs) that are over-complicated, and suboptimal battery management systems. It is logical to measure batteries at frequencies reflecting their daily or weekly charge-discharge cycles, i.e., of the order of microhertz. This is generally not done, however, because of difficulties that include extreme measurement durations and the need for care to avoid issues such as charge distribution problems, charge drift, and the risk of overcharging or flattening the battery. This research demonstrates the feasibility of extra-low frequency (ELF) EIS measurement using standard, non-specialised measurement equipment, and the optimisation of frequency domain data through the superimposition of small-signal measurement tones on larger square wave currents.

Study of charge movement rates in the frequency domain and voltage responses in the time domain in batteries indicates connections between voltage sweep rate in cyclic voltammetry (CV) and current magnitude in incremental capacity analysis (ICA). The key factor determining what the investigator sees, e.g., reversibility of electrode processes or evidence of individual electrochemical reactions, is rate of movement of charge. Thus, controlled current can be used to obtain CV-type data from a battery, something that would be hazardous if done conventionally using a voltage ramp.

Repeatable, low-error EIS measurements at ELFs are essential for inferring battery ECM components, including constant phase elements (CPEs). The fractional order of a CPE is linked to battery state of health (SoH) and energy efficiency. This thesis shows that efficiency measurements with waveforms representative of real battery usage can be used to track battery SoH accurately.