Negative self-potential anomalies can be generated at the ground surface
by ore bodies and ground water contaminated with organic compounds.
These anomalies are connected to the distribution of the redox potential
of the ground water. To study the relationship between redox and
self-potential anomalies, a controlled sandbox experiment was performed.
We used a metallic iron bar inserted in the left-hand side of a thin
Plexiglas sandbox filled with a calibrated sand infiltrated by an
electrolyte. The self-potential signals were measured at the surface of
the tank (at different time lapses) using a pair of non-polarizing
electrodes. The self-potential, the redox potential, and the pH were
also measured inside the tank on a regular grid at the end of the
experiment. The self-potential distribution sampled after six weeks
presents a strong negative anomaly in the vicinity of the top part of
the iron bar with a peak amplitude of -82 mV. The resulting
distributions of the pH, redox, and self-potentials were interpreted in
terms of a geobattery model combined with a description of the
electrochemical mechanisms and reactions occurring at the surface of the
iron bar. The corrosion of iron yields the formation of a resistive
crust of fougerite at the surface of the bar. The corrosion modifies
both the pH and the redox potential in the vicinity of the iron bar. The
distribution of the self-potential is solved with Poisson's equation
with a source term given by the divergence of a source current density
at the surface of the bar. In turn, this current density is related to
the distribution of the redox potential and electrical resistivity in
the vicinity of the iron bar. A least-squares inversion method of the
self-potential data, using a 2D finite difference simulation of the
forward problem, was developed to retrieve the distribution of the redox
potential.