**4. Discussion**

GR compounds are metastable with respect to magnetite [19,42,43]. For instance, depending on pH and dissolved Fe2+ concentration, the precipitation of Fe2+ and Fe3+ can yield either GR(SO4<sup>2</sup>−) or the two-phase system Fe3O4 + Fe(OH)2 [19]. GR compounds are also likely to be spontaneously transformed into magnetite. GR(CO3<sup>2</sup>−) was observed to transform spontaneously under anoxic conditions, resulting in either a mixture of magnetite and siderite (FeCO3) [42] or a mixture of magnetite, chukanovite (Fe2(OH)2CO3) and siderite [43], depending on the pH and the concentration of the carbonate species. The metastability of GR(SO4<sup>2</sup>−) with respect to magnetite explains why the ageing of GR(SO4<sup>2</sup>−) in the experimental abiotic conditions considered here induced the formation of a small proportion of Fe3O4.

However, it was demonstrated that the adsorption of phosphate ions on the lateral sides of the GR(CO3<sup>2</sup>−) particles could prevent their transformation to magnetite [44]. Similarly, lactate ions proved to have a strong effect during the oxidation of GR(SO4<sup>2</sup>−), which was attributed to the adsorption of the lactate ions, through their carboxyl group, on the surface of the GR crystals [37]. It can therefore be proposed that acetate ions adsorb similarly on the lateral sides of the GR(SO4 <sup>2</sup>−) hexagonal platelets (Figure 8) and hinder the formation of magnetite during ageing. The organic polymeric substances associated with bacteria may therefore have similar effects. The bioreduction of lepidocrocite γ-FeOOH by *Shewanella putrefaciens* was studied, and it was observed that the formation of GR compounds was favored, with respect to the formation of magnetite, in the presence of polyacrylic acid or polyacrylamide [45]. Polyacrylic acid, which can model extracellular polymeric substances found in biofilms, was also observed to inhibit, albeit moderately, the reactivity of GR compounds towards methyl red [46]. This reactivity was assumed to mainly involve the Fe(II) reactive sites present on the lateral sides of the GR crystals. Polyacrylic acid, like acetate ions, carries a negatively charged carboxyl group. This confirms that GR compounds can be stabilized by carboxylates.

**Figure 8.** Schematic representation of a GR(SO4 <sup>2</sup>−) crystal with adsorbed acetate ions. The interlayers are actually composed of two layers of SO4 2− ions and include water molecules that were omitted for clarity. The octahedra in green are those built on Fe(III) cations.

Our results showed that the presence of bacteria cells led to stronger effects, because it prevented the formation of magnetite even after 2 months of ageing in anoxic conditions. This effect was the same whether the bacteria were dead (*Pseudoalteromonas* IIIA004) or alive and active (*Micrococcus* IVA008 and *Bacillus* IVA016). The only difference observed was the formation of a very small amount of GR(CO3 <sup>2</sup>−), which resulted from the presence of NaHCO3 in the culture medium and/or the oxidation of organic substances to carbonate by the microorganisms. In the study of the reactivity of GR compounds toward methyl red, it was demonstrated that bacterial cells had a stronger effect on the reactivity of GR(SO4 <sup>2</sup>−) than polyacrylic acid [46]. This is fully consistent with what was observed here by comparing the results obtained in the presence of bacteria and those obtained with acetate. The stronger effect of bacterial cells and/or associated organic species, in particular with respect to the much smaller CH3COO− ions, may have been due to a higher steric effect that would more efficiently hinder the interaction between the GR crystal surfaces (in particular lateral sides) and solution, as illustrated schematically in Figure 8. The transformation of GR(SO4 <sup>2</sup>−) to magnetite requires the release in solution of the SO4 2− ions present in the interlayers of the GR structure, which was assumed to be hindered by the adsorption of anionic species on the lateral sides of the crystals [44]. Voluminous adsorbed species obviously induce a stronger barrier effect than small ones.

The presence of the bacteria, and to a lesser extent, of acetate ions, also led to the absence of GR(Cl−) after 1 week of ageing. Two mechanisms can be proposed to explain this. First, it can be proposed that the bacterial cells and/or the associated organic species accelerate the transformation of GR(Cl−) to GR(SO4 <sup>2</sup>−). However, since bacteria cells tend to stabilize GR crystals, preventing their transformation, this first assumption seems unlikely. The second hypothesis is that bacterial cells and/or associated organic species inhibit the formation of GR(Cl−). During the precipitation reaction, all of the anionic species present in solution may initially adsorb on the Fe-hydroxide sheets that constitute the basic elements of the GR structure [18] (see also Figure 8). In the reference abiotic experiment, only Cl− ions could compete with SO4 2− to adsorb on these hydroxides sheets, which led to the formation of a small amount of GR(Cl−), together with GR(SO4 <sup>2</sup>−). Over time, GR(Cl−) transformed spontaneously to the more stable GR(SO4 <sup>2</sup>−) [23]. When bacteria or acetate were added to the reactants, other species could compete with Cl− and SO4 2−. It can therefore be postulated, in particular with bacteria and the numerous associated organic species, that the preferential adsorption of monovalent organic anionic species on the hydroxide sheets prevented the formation of GR(Cl−) and favored the formation of GR(SO4 <sup>2</sup>−), with SO4 2 being the main divalent available anionic species.
