Batteries, essential components of electric vehicles, are the focus of intense research to improve safety, performance, and fast-charging capability. In this context, measuring the potential of each electrode individually is crucial, not only for laboratory characterisation, but also for intelligent management during commercial operation: in particular, monitoring the negative electrode (NE) potential is a direct and effective way to prevent lithium plating under operando conditions [1]. The reference electrode (RE) provides a practical solution for this, and LiFePO₄ (LFP) REs are among the most promising [2]. Nevertheless, obtaining reliable, reproducible, and artefact-free measurements without perturbing normal cell behaviour remains challenging [3]. This work investigates the impact of REs on cell operation, their electrochemical behaviour, and the accuracy of their measurements.

Six NMC811-graphite cell designs were produced to quantitatively compare LFP and lithium REs. Two dedicated designs, in which the RE covered the full active surface of the cell, allowed a detailed investigation of the underlying operating mechanism of the RE and its impact on cell performance. Pronounced disturbances, differing between lithium and LFP REs, revealed distinct mechanisms underlying the observed artefacts. Similar disturbances were also found in cells equipped with standard REs.

However, the measurements exhibited contradictory trends between polarisation and potential-response delay, which could not be explained based on existing literature. To move towards a more fundamental understanding of the phenomena involved, a pseudo-3D Newman model including a LFP RE was simulated using COMSOL Multiphysics. It reproduced the experimental observations, clarified the origin of these apparent contradictions and revealed a competition between polarisation and delay, providing new insight into RE-related artefacts. This analysis enabled the disturbances to be associated with characteristic deviations in the probed potentials.

Moreover, the simulation showed that REs based on insertion materials behave fundamentally differently from what is commonly assumed. Previously unreported artefacts linked to this phenomenon have been uncovered, raising questions regarding measurement reliability, particularly as the RE ages.

All these artefacts may either accumulate or compensate each other, and their magnitude depends on parameters such as temperature and C-rate. Fortunately, they can be considerably reduced by optimising the RE geometry, dimensions, or initial state. Building on these findings, this presentation will provide practical guidelines for designing artefact-minimised REs, enabling more reliable electrode-resolved measurements in both research and commercial applications.