On the time-dependent electrolyte Seebeck effect

TL;DR

This paper derives the time-dependent electrolyte Seebeck coefficient considering local charge neutrality, and experimentally measures it in various electrolytes, revealing larger values than previously predicted.

Contribution

It provides a new derivation of the time-dependent Seebeck coefficient that accounts for local electroneutrality and validates it through experiments with different electrolytes.

Findings

Measured steady-state Seebeck coefficients around 2 mV/K, larger than literature predictions.

Fitted model to experimental data, obtaining larger ion Soret coefficients than literature values.

Demonstrated the importance of boundary effects and local charge neutrality in electrolyte thermoelectric phenomena.

Abstract

Single-ion Soret coefficients $\alpha_{i}$ characterize the tendency of ions in an electrolyte solution to move in a thermal gradient. When these coefficients differ between cations and anions, an electric field can be generated. For this so-called electrolyte Seebeck effect to occur, the different thermodiffusive fluxes need to be blocked by boundaries -- electrodes, for example. Local charge neutrality is then broken in the Debye-length vicinity of the electrodes. Confusingly, many authors point to these regions as the source of the thermoelectric field yet ignore them in derivations of the time-dependent Seebeck coefficient $S(t)$, giving a false impression that the electrolyte Seebeck effect is purely a bulk phenomenon. Without enforcing local electroneutrality, we derive $S(t)$ generated by a binary electrolyte with arbitrary ionic valencies subject to a time-dependent thermal gradient. Next, we experimentally measure $S(t)$ for five acids, bases, and salts near titanium electrodes. For the steady state we find $S\approx2~\mathrm{mV~K}^{-1}$ for many electrolytes, roughly one order of magnitude larger than predictions based on literature $\alpha_{i}$. We fit our expression for $S(t)$ to the experimental data, treating the $\alpha_{i}$ as fit parameters, and also find larger-than-literature values, accordingly.