Transient oxygen-driven microbial activity at the bentonite–host rock boundary

In many deep geological repository (DGR) concepts for the storage of high-level radioactive waste, bentonite clay is planned to serve as a buffer material between the host rock and the canister. It is expected to serve both a chemical role (immobilization of radionuclides) and a biological one, the inhibition of microbial growth and activity. Sulfate reducing bacteria (SRB) are of particular concern due to the production of sulfide, a strong steel-corroding agent. It is conventionally assumed that microbial sulfate reduction in the bentonite buffer may become active only once oxygen in a DGR is depleted, and that at that stage, compacted bentonite will physically inhibit microbial activity. Here, we challenge this view by showing that a subset of SRB present in the bentonite backfill and in the host rock under repository-relevant conditions are adapted to tolerate, and transiently exploit, oxygen and pore space before the backfill is fully anoxic and saturated. These results are from an in-situ incubation experiment (1.5 and 3 years) conducted in a borehole in Opalinus Clay using modules filled with compacted Wyoming bentonite (1.25 g/cm³) and containing carbon steel coupons. To investigate the role of oxygen, bentonite was pre-equilibrated with an atmosphere containing 0%, 21%, or 100% O₂ prior to deployment. Mineralogical and chemical analyses of the buffer were combined with corrosion studies and microbial assays to assess the response of SRB to oxygen and its consequences. We find that oxygen drives the enrichment of bacteria, including SRB, at the bentonite–host rock interface, most likely during early saturation, when oxygen is still present and the pore space allows for microbial colonization from the borehole. This enrichment leads to sulfide production and reduction of structural Fe(III) in montmorillonite, locally affecting buffer composition, but has a negligible impact on carbon steel corrosion. In the long term (here, 3 years), the oxygen-stimulated effect becomes less important for microbial abundance, which declines; however, sulfate reduction at the boundary remains active. These findings provide a more realistic view of early-stage microbial dynamics at the host rock–backfill boundary and their limited but non-negligible impact on buffer and canister stability in the presence of unavoidable initial oxygen.