Very scarce data are available to build up an inventory of microalgal oil extraction. Char­acterization of the lipid content of microalgae is based on techniques and solvents that cannot be extrapolated to industrial-scale techniques, and often the characterization is done on ly — ophilized algae, which of course is not an option for bioenergy production. Hence inventories for oil extraction and methylester production are usually based on inventories of vegetable oil production and transesterification (e. g., rapeseed or soybean). Some studies specify a phase of pretreatment based on homogenizers. The rapid compression and decompression of the algal slurry is supposed to disrupt cell walls and hence increase extraction efficiency and digest­ibility of extraction residues (Stephenson et al., 2010; Clarens et al., 2011). Triglycerides are extracted with an organic solvent; hexane, lipids, and aqueous phases are then separated and the oil/hexane mixture is finally purified by distillation. During distillation, most of the hexane is recovered, so only a small quantity is lost by volatilization.

In all the concerned studies, triglyceride esterification is performed by reaction with meth­anol and with alkaline catalysis. This step requires heating, mixing, and the addition of a base, usually potassium hydroxide. The reaction yield can be significantly reduced by a concurrent saponification reaction, which is enhanced by water. Consequently there is a trade-off be­tween the energy to invest for dewatering and drying the biomass and the energy for extracting and down-processing the lipid fraction, with reaction yields drastically affected by the water content. Other approaches have been proposed, such as supercritical CO2 extrac­tion of lipids or in situ esterification. Both approaches could suffer from too high water con­tent. More recently, in situ esterification with supercritical methanol has been proposed as a way to overcome this issue. This last option was selected in the LCA-based optimization pro­posed Brentner et al. (2011).