Résumé
The formation of abyssal serpentinites leads to deep changes of the oceanic lithosphere rheology and geochemistry, hence playing a key role on geodynamic and geochemical cycles. Here we review and discuss the geochemical diversity of serpentinites collected on abyssal floors (i.e., abyssal serpentinites) from different geodynamical settings, namely passive margin, forearc and oceanic spreading ridges. We further divided abyssal serpentinites from spreading ridges according to the tectonic contexts in which they were exhumed, differentiating slow-spreading centres (exhumed within the axial valleys and at oceanic detachments), ultra-slow spreading centres (from amagmatic axial valleys and smooth seafloor), fast spreading axis (mostly from deeps) and large transform faults (all spreading rates).The major and trace element composition of abyssal serpentinites is first controlled by melt extraction and melt/rock interaction processes occurring prior to serpentinization. Slow-, fast-spreading ridges and forearc serpentinites are distinguished by low Al$_2$O$_3$/SiO$_2$ ratios and depleted REE signatures when compared to abyssal serpentinites recovered from ultra-slow spreading ridges, transform faults and passive margins, where magmatic extraction is likely more limited or/and melt/rock reaction prominent. Ultra-slow spreading ridge serpentinites have high Fe$^{3+}$/∑Fe when compared to passive margin and forearc serpentinites, while slow spreading ridge and transform fault serpentinites have intermediate Fe$^{3+}$/∑Fe, close to that of magnetite. This distribution is correlated with MgO/SiO$_2$ ratios, suggesting that high MgO contents could thwart Fe oxidation, and thus H$_2$ production, in abyssal environments. The presence of mafic units at depth affects the chemical properties (e.g., sulfur activity [aH$_2$S], oxygen fugacity (fO$_2$)) of the serpentinizing fluids leading to contrasting enrichments of redox sensitive elements (S, U, Eu, Ce, As, Sb) and metals (Zn, Cu) in abyssal serpentinites. At slow-spreading ridges, the circulation of high aH$_2$S and low fO$_2$ fluids, equilibrated with gabbroic rocks, leads to the formation of serpentinites with pronounced Eu anomalies, enrichments of LREE over M-HREE and of As over Sb, and to the storage of metal (Zn, Cu) and sulfide in serpentinites. In contrast, at ultra-slow spreading ridges and at transform faults, the absence of an active magmatic system limits changes in seawater composition prior to serpentinization. The oxidizing conditions favours Ce(IV) (Ce anomalies on REE patterns), and the preferential mobility of Sb(V) over Sb(III) (coupled behaviour between Sb and As), as well as the storage of sulfate over sulfides in rocks. Fast spreading ridge serpentinites present both features with samples having negative Ce anomaly (Ce (IV)) and others with positive Eu* (Eu (II)). Fluid mobile (FME) and non-redox sensitive elements (e.g., Cs, Ba, Rb, B or Li) in abyssal serpentinites are homogeneous at oceanic spreading ridges. Both passive margin and forearc serpentinites are free of Ce anomalies and only few forearc samples display Eu anomalies with moderate LREE enrichments. The forearc serpentinites appear as a geochemical endmember. There, the influence of slab-derived fluids enhances the formation of FME-rich serpentinites having a strong affinity with CO$_2$ -rich sediment-derived fluids (i.e., enrichments in Cs and As relative to Rb and Sb respectively).