MCFA separation

Two separation techniques, liquid-liquid extraction and precipitation, were tested for their capacity to selectively remove caproate and caprylate from the effluent of reactor 2. Effluent was collected throughout the whole fermentation experiment. The effluent contained per liter 9.4 g caproic acid

and 0.26 g caprylic acid and had a pH of 6.52. The pH determines the efficiency of each separation technique; extraction only separates undissociated acids from the water phase, whereas precipitation only occurs with the dissociated form. Therefore, the separation tests were performed with the fermentation medium at pH 5, 6.52 and 7, which was adjusted by titrating with 2 M NaOH or HCl solutions.

Extraction

Liquid-liquid extraction is a separation technique in which the target component is transported from one liquid, namely the fermentation broth, into another liquid, specifically a solvent that can be removed easily. Two different solvents, ethyl hexanoate and petroleum ether, were tested for their suitability to selectively remove caproate and caprylate from the fermentation broth by extraction. The extraction was performed in 50 mL separation funnels of glass (Schott, Germany), which were washed 3 times with acetone before usage. After filling the funnels with 27 mL of fermentation broth and 6 mL of solvent, they were shaken for 15 min at 150 rpm. After shaking, the funnels were left for 15 minutes to separate into two phases. Then 1 mL samples were taken from the fermentation broth to analyze the residual VFA concentration. All extractions were performed in duplicate.

Distribution coefficient calculations

After extraction, the undissociated acid in the solvent phase [HA]s is in equilibrium with the undissociated acid in the water phase [HA]w. The distribution of the undissociated acid among both phases is expressed in the distribution coefficient. We calculated the effective distribution coefficient based on the total acid concentrations as in Equation 6.1. The solvent concentration (HAsTot) is calculated from the difference in acid concentration before (initial) and after the extraction per volume fermentation broth (Vw), divided by the solvent volume (Vs). The effective distribution coefficient was calculated for each extraction with initial pH values of 5, 6.52 and 7.

[ ]

[ ]

HA

HA

D=

[

]

[

]

[

]

V

HA

HA

V

=

HA

Activity test in extracted medium

After extraction, the fermentation medium was tested for the ability to act as a substrate again for bacterial growth. Residues of the solvent in the medium could be toxic for bacteria and thus impede fermentation with recycled medium. Bottles were filled with the extract and with non-extracted fermentation medium (control), sealed with rubber inlets and capped with aluminum crimp caps. After purging the headspace 8 times with pure hydrogen to a final pressure of 1.5 bar, 1 mL of inoculum was injected anaerobically. At day 0 and day 58, the pressure of the headspace was measured and a liquid sample was taken to analyze the VFA and alcohol concentrations.

Precipitation

The separation technique precipitation relies on differences in solubility products, with the requirement that the target component has a lower solubility product compared to other components in the system. Precipitation with a divalent cation was preferred over a monovalent cation in order to reduce the amount of salt addition and to maintain a lower solubility product. Of the common divalent cations, copper, magnesium, barium and calcium, calcium was selected as ligand for the precipitation of caproate and caprylate. Calcium, like the divalent cation copper, had a lower solubility product with caprylate than Mg and Ba salt (Papageorgiou, unpublished results), and, unlike copper, is not a highly toxic ion.

Precipitation was performed with fermentation broth at pH 5, 6.52 and 7 with different amounts of calcium to estimate the solubility product and the amount of calcium needed to separate caproate and caprylate as calcium carboxylates from the fermentation broth. To a 100 mL bottle containing 50 mL of fermentation broth. 0.05, 0.1, 0.5, or 1 g of calcium chloride dihydrate was added. Additionally, in the case of the fermentation broth at an unaltered pH 6.52, experiments adding 2, 3 and 4 g were performed. After gentle mixing, the bottles were kept in a 20ºC room for 15 hours. After the suspension was filtrated by gravity, the filtrates were analyzed for residual FA concentration. Before and after precipitation experiments, the pH of solutions and filtrate was measured.

6.2.4 Analysis

Hydrogen, oxygen and methane were analyzed by gas chromatography as previously described (Steinbusch et al., 2008). Fatty acids (C2-C8) were analyzed by gas chromatography using a HP 5890 serie II with a glass column (2m x 6mm x 2mm) packed with 10% Fluorad 431 on Supelco-port 100- 120 mesh. Prepared samples (1.0 µl) were injected directly on the column at 200˚C. The carrier gas was 40 mL min-1 _{nitrogen saturated with} formic acid. Oven temperature was 130°C for 4 min, raised at 12°C min-1 ramp to 160ºC for 6.5 min. Fatty acids were detected with a FID at 280˚C using hydrogen at 30 mL min-1 _{and air at 400 mL min}-1_{. Alcohols (C}

2-C6) were measured using the same column, but at a constant oven temperature of 70ºC.

Protein concentration determination with modified Hartree-Lowry method (Caprette, 1995) was used to quantify bacterial cell concentration in a small sample volume. One mL of reactor liquid was transferred in a 2 mL vial and centrifuged for 5 min at 10.000 rpm. The supernatant was removed and the pellet was resuspended in 1 mL of 1 M NaOH. The vial was left for 30 min at 50ºC in a water bath to hydrolyze cells. Subsequently, protein concentration of the hydrolyzed cells was quantified using modified Hartree-Lowry method. It was experimentally found that 1.0 g of protein equals 3.0 g of volatile suspended solids (VSS).

6.3

Results and discussion