The original iron precipitate in the BIFs was probably ferrihydrite, Fe2O3·xH2O. The ferrihydrite was precipitated out of equilibrium with the atmosphere by oxidation of upwelled ferrous iron, either by O2 produced by cyanobacteria within the water column or by phototrophic, anoxygenic, iron-oxidizing bacteria. Once in the sediment, ferrihydrite was subsequently converted to a stable iron oxide, either magnetite or haematite, or to the reduced mineral siderite. The constraint on derives from applying the following equilibrium reaction between magnetite and siderite: . As discussed below, is a measure of the redox potential of the system, but organic matter (CH2O) was the reducing agent in this reaction.

In the model of Rosing et al.4, was controlled by methanogens. Anaerobic ecosystems of this nature have been studied by ref. 5. The downward flux of H2 through the atmosphere–ocean interface is limited by its piston velocity to (1–6) × 1011 molecules cm−2 s−1 (Table 2 in ref. 5). The average deposition rate of Fe3O4 in BIFs is estimated to be 0.1–1 mm yr−1, or (40–400) × 1011 molecules cm−2 s−1, assuming that each microband represents one year of deposition6. Suggestions that the deposition rate was much slower than this7,8 are probably biased by hiatuses in the geologic record. Reducing this iron to siderite would require an equal flux of H2, which is 7–400 times the downward H2 flux into the ocean estimated above. Some H2 could have been produced within the sediment via fermentation, but this requires that organic matter be available. Formation of siderite without a reducing agent—that is, by disproportionation ()—would require an effective of 3 × 10−6 (the value at the magnetite–haematite boundary), which is ten times smaller than the value that Rosing et al.4 propose and is probably impossible to achieve.

Probably the reducing agent was not H2 but organic matter produced by photosynthesis in the water column. A fraction of the primary production of organic matter was exported from the photic zone to the sediment, where microbial iron respiration could produce siderite9,10,11,12: . Light carbon isotope ratios are found in siderite, but not in associated Ca/Mg carbonates11,12,13, indicating that a portion of the carbon in siderite was indeed derived from a pool of isotopically light organic matter. In BIFs, magnetite was initially in disequilibrium with the overlying CO2-rich atmosphere and was converted to siderite. However, the above reaction was only allowed to proceed as long as a reductant (organic matter) was available. The rain of organic carbon into the sediment was not sufficient to keep pace with that of Fe3+-bearing oxide, and its exhaustion led to a mineral assemblage (coexisting magnetite and siderite) that did not reflect equilibrium conditions with the atmosphere–ocean system. The transition from siderite to iron oxides in BIFs from the Campbellrand–Kuruman complex could have been caused by the decreased supply of organic material in offshore regions10.

To summarize, microbial cycling and diagenesis of organic matter and ferric iron controlled the effective and in the sediment. The mineralogy of BIFs reflects these conditions, with no simple relationship to atmospheric CO2. Thus, in the atmosphere may have been high enough (about 0.1 bar) to explain the warm Archean climate, without the need for additional warming mechanisms.