Investigating the electrode-electrolyte interface modelling in cochlear implants

Proposing a good electrode-electrolyte interface (EEI) model and properly identifying relevant parameters may help designing safer and more optimized auditory nerve fiber stimulation and recording in cochlear implants (CI). However, in the literature EEI model parameter values exhibit large variability. We aim to explain some root causes of this variability using the Cole model and its simpler form, the Basic RC model.

We use temporal and spectral methods and fit the models to stimulation pulse voltage response (SPVR) and electrochemical impedance spectroscopy (EIS) data.

Temporal fittings show that there are multiple sets of model parameters that provide a good fit to the SPVR data. Therefore, small methodological differences in literature may result in different model fits. While these models share similar characteristics at high frequencies >500 Hz, the SPVR fitting is blind to low frequencies, thus it cannot correctly estimate the Faradaic resistor. Similarly, the polarization capacitor and its fractional order are not estimated robustly (capacitor variations in the nano- to micro-farad range) due to limited observation of mid-range frequencies. EIS provides a good model fit down to ~3Hz, and thus robust estimation for the polarization capacitor. At lower frequencies charge mechanisms may modify the EEI, requiring multi-compartment Cole model fitting to EIS to improve the estimation of Faradaic characteristics. Our EIS data measurements down to 0.05Hz show that a two-compartment Cole model is sufficient to explain the data.

Our study describes the scope and limitation of SPVR and EIS fitting methods, by which literature variability is explained among CI EEI models. The estimation of mid-to-low-frequency characteristics of the CI EEI is not in the scope of the SPVR method. EIS provides a better fit; however, its results should not be extrapolated to unobserved frequencies where new charge transfer mechanisms may emerge at the EEI.