Archean cherts potentially constrain the primitive environment in which life emerged and evolved. These siliceous rocks formed by three processes : C-cherts (primary cherts) formed by the chemical precipitation of oceanic silica, either as a siliceous ooze (or silica gel) on the seabed, or as cement within still soft sediments at the surface ; F-cherts (fracturefilling cherts) precipitated from circulating fluids in concordant or crosscutting veins in the shallow crust ; S-cherts (secondary cherts) are the result of the metasomatism (silicification) of preexisting rocks during the percolation of silica-rich fluids. These processes are generally accepted but major questions remain unsolved : how to recognize various chert types ? Where does the silica come from and how did it precipitate ? What chemical signal is hosted in cherts and how can it be used for paleo environmental reconstructions ? These questions are addressed here using three sites in the Barberton Greenstone Belt, South Africa, which contain a variety of cherts deposited in very different environments. The approach combines field description of sedimentary and deformation structures, the characterization of various chert petrologies, and the study of their chemical and isotopic composition. In these three sites, chert formation strongly depends on the environmental setting. Clastic sedimentation is directly linked to C-chert formation at Komati River, where the silica was deposited as a viscous, siliceous ooze by sorption process onto suspended clay particles. A continental contribution is absent at Buck Reef, and the black and white banded cherts (C-cherts) are interpreted to have formed by chemical precipitation of oceanic silica during seasonal (black chert) and maybe glacial/inter-glacial (white chert) climatic variations. The fracture-filling cherts from Barite Valley precipitated from a thixotropic colloidal suspension that migrated upward through the crust. The chemical compositions of cherts from these three sites are essentially controlled by the environment of deposition, and represent mixtures of a siliceous and contaminant phases, independent from the silica precipitation mode. Komati River C-cherts and Barite Valley F-cherts are both enriched in Al, K, Ti, HFSE and REE, which represents the contamination by phyllosilicates of the microquartzitic fabrics. Such a clastic contribution may account for the wide range of δ30Si in Komati River cherts (-0.69‰ to +3.89‰) although the majority of positive values is attributed to seawater involvement. In the dykes, δ30Si is strongly negative (-4.5‰to +0.22‰) and is consistent with the low-temperature hydrothermal nature of these fluids. At Buck Reef, the lack of continental contribution is expressed in the white cherts, by a mineralogy exclusively composed of microquartz, and by extremely low trace element contents, i.e. HFSE and REE below 1ppm. We calculate that 2% of carbonates and 3-4% of clastic particles (i.e. clay, feldspar) would be enough to mask the silica composition in these high purity cherts. A marine signature was not recognized in their geochemistry because of the unreliability of commonly used modern proxys (i.e. LREE depletion, La and Y enrichment). These features were identified in oceanic, hydrothermal, magmatic and pegmatitic quartz and thus do not reliably identify an oceanic signal in Archean cherts. Because the δ 18O values in these white cherts indicates secondary fluid circulations at <100°C, their negative or positive δ30Si values (-2.23‰ and +1.13‰ in average) most probably represent different fluid contributions at the time they formed. The combination of field and petrological observations appears to be the most reliable approach to classify cherts and to deduce their origin, and we show here that their chemical composition depends more on the environmental conditions than on the primary fluid characteristics.
