(275c) Intensifying the Recovery of Carboxylic Acids from Fermentation Broths Via Reactive Extraction inside the Membrane-Assisted Spinning Disc Reactor

The production of carboxylic acids by fermentation has gained research attention to achieve a renewable option as compared to fossil carbon-based benchmarks. A major bottleneck lies on the recovery from fermentation broths of these acids, which are normally present at dilute concentration (ca. 5 wt.%). Reactive extraction is an alternative to solvent extraction. The process involves a reaction between an extractant, a diluent, and the carboxylic acid. The extractant, often a tertiary amine, form complexes with the acid that are soluble in the organic phase. Reactive extraction of carboxylic acids outperforms conventional solvent extraction, due to its superior partition coefficients obtained when selecting a suitable extractant and diluent. Most reactive extraction studies has been carried out in batch systems at relatively low stirring speeds (< 150 RPM), where the reaction takes place in the liquid-liquid interfacial area. In this concept, vigorous stirring is preferred to quickly reach equilibrium. However, this leads to the formation of un-wanted, stable emulsions which often require a subsequent centrifugation stage. Full phase separation may take up to 36 hours after extraction in conventional batch systems at 500 RPM, which represents a limitation for the continuous process (see Attachment 1). In this work, process intensification of reactive extraction has been carried out inside a novel reactor, the membrane-assisted spinning disc reactor (MASDR), to recover diluted concentrations (ca. 5 wt.%) of propionic acid (PA), which is a valuable chemical used as preservative in the food industry. The concept is depicted in Attachment 2.

Here, the solvent and the feed are co-fed into the MASDR, where high shear is ensured to reach equilibrium conditions very fast. However, above a certain shear rate very stable emulsions would form, which would take very long times to phase-separate. By placing a membrane at the bottom side of the spinning disc, phase separation may be achieved by breaking the stability of emulsions due to forced transport through porous media, ultimately reaching demulsification. Extraction and phase separation may be thus achieved in the same equipment, making it promising for biotechnological applications.

Preliminary equilibrium studies for the extraction of propionic acid using pure solvents (i.e., solvent extraction), or combinations of a solvent with a tertiary amine (i.e. dimethyldodecylamine - DDA - or tri-n-octylamine - TOA -), were carried out in order to find the best extractive system for propionic acid recovery from synthetic fermentation broths (< 4.5 wt. %). The results of the equilibrium studies are presented in Attachment 3. The extractive system TOA in 1-Decanol was selected as the desired solvent for reactive extraction, and after having studied the effect of TOA/PA molar ratio on extraction efficiency, it was inferred that PA extractions above 75% can be achieved if only TOA/PA mol ratios are higher than 0.58 (see Attachment 4).

Batch and MASDR experiments were carried out at 35 ËšC for the recovery of propionic acid (0.6 mol/L), using a mixture of 30 vol.% TOA in 1-decanol. The aqueous-to-organic ratio (volume/volume) was two. The results are shown in Attachment 5.

In Attachment 5, it was concluded that the high shear induced by MASDR helped to quickly achieve equilibrium from 200 RPM onwards (i.e. droplet-droplet emulsion region), which reflects its superior performance in terms of mixing compared to batch systems. However, going above 500 RPM did not improve performance any further, even at residence times as low as 24 seconds. In fact, it should be prevented as higher shear rates generate much more stable emulsions which are extremely hard to separate. If these stable emulsions, once formed, were collected and re-entered the MASDR one more time at e.g. residence time of 1 min, partial demulsification can be quickly achieved after 10 min, as shown in Attachment 6. Phase separation is clearly visible, which does not occur in long-time ordinary settling, though full separation was still not reached for that timespan. By limiting MASDR operation below 500 RPM, the extent of demulsification may be much higher, optimizing the process. Further study on demulsification properties under those operating conditions is currently in progress, and a more detailed explanation will be given at the presentation.