LETTER TO THE EDITOR Unfounded Solvent Recovery Process by Membrane Filtration Albert J. Dijkstra Received: 3 August 2011 / Revised: 19 September 2011 / Accepted: 21 September 2011 / Published online: 9 February 2012 Ó AOCS 2012 Dear Editor, In my opinion, the article [1] contains many misunder- standings, inaccuracies, misleading statements and unfoun- ded conclusions. Accordingly, it requires some rectification. The authors report results of experiments dealing with solvent recovery from miscella by membrane filtration and compare this strategy with conventional distillation pro- cesses. The authors claim that conventional solvent recovery by distillation consumes about 50% of the total energy in edible oil extraction but they forget that most of this energy is supplied by the vapors leaving the desol- ventizer/toaster and thus there is little net energy saving. The full miscella has an oil content of some 25% which is equivalent to three parts of solvent per part of oil [2] and the first stage evaporator increases the oil content of the miscella to some 75%. Only 1/3 part of solvent per part of oil and only the evaporation of this 1/3 part requires steam. Accordingly, the authors are off by a factor of eight in estimating potential energy savings. In fact, the authors totally disregard marc desolventi- zation. They disregard the fact that the solvents they studied have higher latent heats of evaporation than hexane and, as opposed to hexane are miscible with water. Accordingly the condensate from the desolventizer/toaster cannot be sent to the equivalent of a hexane/water separator but instead must be processed in a fractional distillation. The authors use iso-propanol as extraction solvent and seem to ignore much past research on alternative oilseed extraction solvents [3, 4]. This solvent has been studied intensively [5] since it has the advantage that while hot, it dissolves triglyceride oil but that on cooling, the miscella forms two phases: an oil-rich phase containing some iso- propanol and an alcohol phase containing so little oil that is can be recycled to the extractor as such [3, 4]. So why bother separating the alcohol from the miscella using membranes when cooling will readily achieve what is aimed for with a lower capital investment and energy usage? The authors also refer to a nanofiltration application for the separation of refined soybean oil from a mixture of this oil in liquid n-butane [6]. Why not just release the pressure like people do with carbon dioxide [7]? The authors claim ‘‘huge potential savings when mem- branes are used due to the reduction of chemicals and the improved quality of the oil.’’ Their own experiments do not justify this statement since like Raman et al. [6] they just added free fatty acids to a fully refined oil purchased in a local supermarket. Consequently, the authors do not know how the various phosphatides, sterols, pigments, etc. pres- ent in the crude oil miscella will behave when this miscella is subjected to membrane filtration. Will they foul the membrane? That would most certainly decrease the afore- mentioned ‘‘huge potential savings’’. Also, the removal of phospholipids from miscella is not known as the ‘‘dewaxing process’’as the authors state but rather as ‘‘degumming’’. The authors mention that ‘‘phospholipids can be retained by ultrafiltration (UF) membranes’’ but do not provide a specific literature reference. In this context, they could have referred to the patent by Ko ˜seog ˘lu et al. [9], which shows that in some instances, a membrane can retain non- hydratable phosphatides. However, that does not mean that the membranes used and recommended by the authors are similarly suitable for degumming. Consequently, the sav- ings claimed by the authors lack substantiation. Contrary to what the authors state, cyclohexane is not used industrially for the extraction of ‘‘vegetable seeds such as sunflowers or soybean plants’’. When reviewing A. J. Dijkstra (&) Carbougne `res, 47210 St. Eutrope-de-Born, France e-mail: albert@dijkstra-tucker.be 123 J Am Oil Chem Soc (2012) 89:957–958 DOI 10.1007/s11746-012-2008-z