Geosmin and 2-MIB have been identified as the major taste and odour causing compounds in both drinking water and in recirculating aquaculture systems (RAS). These two compounds are produced by actinomycetes and cyanobacteria as secondary metabolites, resulting in an earthy/musty odour. Although they are not considered harmful to human health, these compounds should be removed to sustain the acceptability of the drinking water and the taste of farmed fish. Humans can detect them at very low concentrations (ng/L).

In RAS, fish absorb geosmin and MIB mainly through the gills, and additionally through the gut and skin. Geosmin and MIB are transported through the blood before deposition in tissues. In general, geosmin absorption is faster than MIB absorption, while excretion of geosmin is slower. Therefore, geosmin bio-accumulates to higher levels in fish and requires more time to be excreted from the fish than MIB. The common practice to remove off-flavour consists in relocating the fish to a biofilm-free depuration flow-through system receiving clean ‘odour- free’ water. Off-flavours delay harvest and sales or make fish unmarketable. To reduce costs for remediation, off-flavours in systems and in fish must be detected and prevented as early as possible.

Conventional water treatment practices (coagulation/flocculation, filtration and disinfection with chlorine) are inefficient in removing these substances. The application of oxidants such as chlorine, chlorine dioxide or potassium permanganate does not provide a complete removal due to the resistance of the tertiary alcohols towards oxidation. The use of biocides to control actinomycetes in RAS is also difficult. The frequent cleaning of the heat exchanger and biofilters and the fast removal of organic matter helps to reduce off-flavour-producing microorganisms in RAS; but it is time consuming. Ozone and the addition of activated carbon can efficiently remove these compounds. However, the adsorption efficiency of activated carbon is reduced due to the presence of natural organic matter (NOM) and its regeneration is time consuming; leaving ozone as the only solution. The auto-decomposition of ozone results in the formation of hydroxyl radicals (·OH). As MIB and geosmin are considered to be susceptible to attack by ·OH, this would ultimately lead to a reduction of both compounds by ·OH. Advanced oxidation processes (AOPs), which involve the formation of highly potent chemical oxidants, such as ·OH, have been studied and exhibited a reduction of taste and odour compounds below the threshold odour concentration. AOP normally combine high capital and energy costs, and the oxidation can produce by-products (DBPs).

The NOM contained in water has a significant effect on the decomposition of ozone and consequently the ozonation of MIB and geosmin. Both compounds are difficult to oxidize with ozone because they consist of saturated ring systems. High pH promoted the formation of ·OH by the auto-decomposition of ozone, and the removal efficiency of both compounds increased as pH increased. These results support the theory that the destruction of MIB and geosmin is primarily through the ·OH. MIB was found to be more resistant to ozonation than geosmin, which may be attributed to the increased number of tertiary carbons in the MIB molecule.

Water ApS and Oxyguard International A/S are participating in a VIDA project: “Saving water by removal of off-flavour in farmed fish”. Preliminary experiments in Water ApS labs have also shown a positive correlation between ozone addition and removal of geosmin and MIB in clean water with limited organic matter. Water samples were spiked with different concentrations of gesomin and MIB (50 ng/L and 500 ng/L). A 90-95% reduction in both geosmin and 2-MIB concentrations was found when treated with ozone. Nonetheless, the understanding of the kinetics of removal geosmin and MIB by ozone is necessary. Thus, more experiments are performed regarding the removal of these compounds by ozone in RAS water in order to design efficient and affordable ozonation systems.