Biological Selenite Reduction and Biofilm Growth in a Microfluidic Flow Cell

Abstract:

Selenite-contaminated groundwater can be biologically remediated in-situ by supplying an electron donor to promote the growth of selenite-reducing bacteria. We studied the fate of selenite during in-situ bioremediation using a microfluidic flow cell containing a homogeneous distribution of pores. The flow cell had two inlets: one for selenite supply, and the other for propionate (electron donor) supply. The media contained sulfate, which is common in groundwater and can affect selenite reduction. During the 5-month operation, biomass and selenite reduction products were periodically imaged using a phase contrast microscope and an environmental scanning electron microscope. Selenite reduction products were further characterized using Raman spectroscopy and energy-dispersive X-ray spectroscopy. Three types of crystals were detected in the mixing zone between selenite and propionate, and they occurred in different locations of the mixing zone. On the selenite side, selenite was biologically reduced to elemental selenium in the monoclinic form. Along the centerline, sulfate was biologically reduced to sulfide, which chemically reacted with selenite to form the second type of crystal (selenium sulfide). On the propionate side, selenium sulfide was biologically reduced to elemental selenium in the trigonal form. A mathematical model was developed to explain the segregation of the three crystals. On the selenite side, bacteria preferred selenite to sulfate since selenite can provide more energy for bacteria growth according to thermodynamics. On the propionate side, selenite was limiting; thus bacteria used selenium sulfide as their electron acceptor. Understanding this segregation can help to predict in-situ bioremediation of selenite-contaminated groundwater. Conventional prediction models consider the reaction on the selenite side as the only path of selenite removal, while a model considering the three paths of selenite removal would increase the prediction accuracy.