**4. Discussion**

The experimental device was designed so that, inside a channel of small dimension filled with seawater, the ohmic drop could be important. As a result, the coupon C12 set the farthest inside the channel had an initial potential, corrected from ohmic drop, equal to *E*IR free = −0.61 V vs. Ag/AgCl-seawater. The applied potential was equal to −1.06 V vs. Ag/AgCl-seawater, so that the ohmic drop was 0.45 V for C12. At the beginning, coupon C12 was not protected. Its potential however decreased during the first month and remained constant at *E*IR free = −0.91 ± 0.01 V vs. Ag/AgCl-seawater from day 32 to day 109. The decrease observed in the first month is the consequence of the formation of a calcareous deposit on the surface of the coupons located close to the channel entry, which were correctly protected as soon as CP was applied. For instance, the initial value of the potential was *E*IR free = −0.95 V vs. Ag/AgCl-seawater for C1. Because of the progressive formation of the calcareous deposit, the ohmic drop decreased with time, allowing more and more coupons to be correctly protected and covered with a calcareous deposit.

At day 28, when the seawater flow inside the channel was interrupted, pH measurements demonstrated that the pH was increasing inside the channel, from C4 to C7 to C9 (Section 3.1). Because the interfacial pH is linked to the cathodic reaction rate, this result shows that the cathodic current density was increasing from C4 to C7 to C9. The already mature calcareous deposit formed on the surface of C4 implied a strongly decreased current demand. In contrast, the potential of C9 had just reached its minimal value approximately at day 28 (see Figure 5), which implies that the current demand was important. The deposit formed on C9 at higher average potential was moreover less "efficient" than the deposit formed on C4.

A steady state could, however, be reached for the whole system, approximately at day 32. This result shows that, once the deposit has formed on each coupon, and reached its "final" state (in terms of thickness and porosity), the current demand from each coupon became constant. The overall current flowing at a given position in the channel remained constant and the ohmic drop could not decrease any further. This means that the potential reached by the coupons located far inside the channel, i.e., C9–C12, could have been modelled through a theoretical computational approach only if the influence of the calcareous deposit forming on each coupon could have been modelled. Even though the modelling of calcareous deposition was already reported [13,14], this seems challenging in the present case because the effects of the calcareous deposit depend on the potential of the metal, which, in this particular problem, depends itself on the properties of the deposit formed previously on other coupons.

Moreover, coupon C6 set in the smaller part of the channel, though following the same trend observed for other studied coupons (C1, C4, C9 and C12), was characterized by two specific features. The first one relates to voltammetry: the cathodic part of the polarization curve for C6 differs from that of the polarization curves of C4 and C9. Secondly, as for the other coupons in the smaller part of the channel, XRD revealed a strong preferential orientation of the aragonite particles constituting the calcareous deposit. It can then be forwarded that the properties of the deposit formed on C6 differed from those of coupons C4 and C9. In the smaller part of the channel, the flow velocity was higher (up to 10 cm s<sup>−</sup><sup>1</sup> according to the measured flow rate), which is known to have an impact on calcareous deposition [4,12].

The efficiency of CP, assessed via voltammetry experiments, was observed to be associated not only to the cathodic level reached by a coupon at a given time, but also to the previous level of CP, that governs the protective efficiency of the calcareous deposit. This last point is also illustrated by the low corrosion rates estimated for coupons C1 and C4 when CP was stopped (i.e., at *E* = *E*cor). This protective ability was already observed to be important for aragonite layers [15]. It may also be partially due to the formation of a passive or pseudo-passive layer at the steel surface, favored by the increase of the interfacial pH. Such a layer would have more likely formed on the coupons polarized at the lowest potentials, and thus typically on coupon C1. This could explain the typical features of the polarization curve of coupon C1 (see Figures 7 and 8), i.e., a slightly different anodic Tafel slope and (ii) a stronger effect on O2 diffusion in the cathodic domain (with respect to C4 and C9).
