A floating sludge sample from the first basin of the wastewater treatment plant of an oleochemical industry (Oleon NV, Ertvelde, Belgium) was used to enrich the specialized glycerol-degrading microbial community. The selection of this community was motivated by the high load of chemical oxygen demand (COD) of 3716 kg COD day⁻¹, mainly from fatty acid synthesis as well as biodiesel production. The sample contained mainly propionate (315 mg COD L⁻¹), acetate (173 mg COD L⁻¹), and glycerol (160 mg COD L⁻¹). Samples were stored at −80°C and revived following the protocol as described by Kerckhof et al. (2014) using cryoprotection agents, such as dimethyl sulfoxide, plus trehalose and tryptic soy broth. Enrichment of glycerol degraders was performed in 20 mL Balch tubes using modified homoacetogenic medium (HA medium) as described by Patil et al. (2015a), supplemented with 15 g L⁻¹ of glycerol as the only carbon source. The medium was adjusted to pH 6.0 with 0.1M phosphate buffer. The tubes with growth medium were inoculated with a suspension of 10⁻³ dilution of original sludge sample, prepared in miliQ water, and incubated at 28°C for an initial period of 11 days. Following three subsequent transfers (10%, v/v) in freshly prepared medium on a weekly basis, the mixed microbial community was then transferred and further enriched at the same incubation conditions for 20 days in 40 mL penicillin bottles using an autoclaved modified growth medium as described by Dennis et al. (2013). Medium composition: 2 g L⁻¹ Na₂HPO₄, 1 g L⁻¹ KH₂PO₄, 1 g L⁻¹ NH₄Cl, 0.0147 g L⁻¹ CaCl₂·2H₂O, 0.1 g L⁻¹ MgSO₄·7H₂O and 15 g L⁻¹ glycerol. The medium was supplemented with 1 mL L⁻¹ of a mixed trace element solution and 1 mL L⁻¹ of a vitamin solution (Patil et al., 2015a). During all enrichment experiments, the incubations were sampled on a weekly basis.

Two independent BESs were constructed using Perspex frames. A two-compartment system (2C) with a cation exchange membrane (CEM) (Fumatech FKB, Fumasep, Germany) fixed between the cathode and anode compartments was considered as control reactor for 1,3-PDO production as previously described (Zhou et al., 2013). A three-compartment system (3C; Gildemyn et al., 2015; Figure 1), dedicated to 1,3-PDO production and in situ organics extraction, was built with an additional AEM (Fumatech FAB, Fumasep, Germany) between the cathode and extraction compartments. Both membranes were pretreated in accordance to the specifications of the company. Acid and base pretreated carbon felt (3 mm thick; 5 × 20 cm exposed to the liquid; Alfa Aesar, Germany) was used as a cathode electrode with a stainless steel frame current collector. An IrOx-coated titanium mesh (5 × 20 cm; Magneto Special Anodes BV, the Netherlands) was used as a stable anode. In the cathode compartment, an Ag/AgCl electrode (3M NaCl, 0.209 V vs. standard hydrogen electrode; RE-1B, Biologic Science Instruments, France) was used as a reference electrode. All electrodes and membranes have a projected surface area of 100 cm². All reactor compartments have a working volume of 350 mL, which includes 150 mL of liquid volume in a closed recirculation buffer vessel of 500 mL. The pH of the cathode was controlled with 2M NaOH and 2M HCl at pH 6.0 (Dulcometer D1Cb/D1Cc, Prominent, Netherlands).

The cathodes were inoculated with a mixed-microbial community enriched with specialized glycerol degraders. The initial concentration of intact cells was set at 4.0 × 10⁷ cells mL⁻¹ as determined by flow cytometry according to De Roy et al. (2012). The catholyte was a modified medium as reported by Dennis et al. (2013). The anolyte was 0.1M H₂SO₄ (pH 2). The medium in the extraction compartment consisted of four times concentrated inorganic catholyte medium. The catholyte, anolyte, and the extraction compartment liquids were recirculated at a rate of 3.6 L h⁻¹ to facilitate mixing. The electrochemical experiments in both BESs were conducted galvanostatically at a fixed current density of 10 A m⁻² using a VSP potentiostat (Biologic, France). The resulting cell potential was around 2.78 ± 0.07 V throughout the whole operating period. The BESs were operated in a fed-batch mode; 15 g L⁻¹ (18.24 g COD L⁻¹) of glycerol (min 97% purity, Laborama VWR BDH Prolabo, Netherlands) was added as soon as the glycerol concentration decreased close to 0 g L⁻¹. The catholyte, extraction medium, and anolyte were sampled three times per week. Total gas production was determined three times a week by means of a water displacement column assembly. The relative gas composition was determined by analyzing a gas sample from the water displacement column.

Organic acids and cathode gas composition were analyzed according to Patil et al. (2015a). Alcohols were analyzed according to Gildemyn et al. (2015). Samples for organic acids and alcohol analysis were stored at −20°C and appropriately diluted with milliQ water before analysis. Organic acids (C1-C5) were analyzed using a 930 Compact IC Flex (Metrohm, Switzerland) ion chromatography system with inline bicarbonate removal (Metrohm CO₂ suppressor). Separation was done on a Metrosep organic acids (250/7.8) column at 35°C behind a Metrosep organic acids (4.6) guard column. The eluent was 1 mM H₂SO₄ and the flow rate 0.5 mL min⁻¹. An 850 IC conductivity detector was used for detection of eluted components. Detection was enhanced using a chemical suppression module to replace protons with Li-cations (Metrohm suppressor module with 0.5M LiCl). The sample aspiration needle was cleaned with acetone between each analysis. The lower limit of quantification was 1 mg L⁻¹ (Gildemyn et al., 2015). Alcohols were analyzed using gas chromatography-flame ionization detector (GC-FID, Varian CP-3800 Gas Chromatograph, USA) during the enrichment phase of the study. The GC-FID has detection limit of 2.0 mg L⁻¹ to 1000 mg L⁻¹. During the BES experiment phase, alcohols (C1-C4) were analyzed using a 930 Compact IC Flex (Metrohm, Switzerland) ion chromatography system. Separation occurred at 35°C on a Metrosep Carb 2 (250/4.0) column behind a Metrosep Trap 1 100/4.0 guard column. The eluent was 20 mM NaOH at a flow rate of 0.8 mL min⁻¹. An IC amperometric detector (cycle: 300 ms 0.05V, 50 ms 0.55 V, 200 ms −0.1 V detection during 200–300 ms of each cycle) was used for detection of eluted components. The sample aspiration needle was cleaned with acetone between each analysis. The lower limit of quantification was 1 mg L⁻¹. Standards and controls were regularly used for calibration of the signals (Patil et al., 2015a). Effluent gases from reactors were analyzed on a Compact GC (Global Analyser Solutions, Breda, the Netherlands), equipped with a Molsieve 5A pre-column and Porabond column (CH₄, O₂, H₂, and N₂) and a Rt-Q-bond pre-column and column (CO₂, N₂O, and H₂S). Concentrations of gases were determined by means of a thermal conductivity detector. The carrier gases were N₂ for H₂ channel and He for all other gases. The detection limit for these gases was 100 ppmv.

Coulombic efficiency (CE) and production rate per electrode surface were calculated as previously described by Patil et al. (2015b). Glycerol can be fermented toward hydrogen, therefore CE reported in this work are based on total charge delivered to the cathode through electrical current and glycerol. For example, the overall CE of 1,3-PDO production is calculated according to Eq. 1.

CE1,3PDO=(([1,3PDOc;t]−[1,3PDOc;t−1])×Vc+([1,3PDOm;t]−[1,3PDOm;t−1])×Vm+([1,3PDOa;t]−[1,3PDOa;t−1])×VaM1,3PDO)×n1,3PDOA((JappF)×3600×24×(t−t−1)+([Glyc;t−1]−[Glyc;t]Mgly)×nglyA)×100 with CE_1,3-PDO as the overall CE of 1,3-PDO production in %, [1,3-PDO_x;y] as the concentration of 1,3-PDO in g L⁻¹ in a certain compartment at a certain time point (time in days), V_x the volume of the corresponding reactor compartment in L, M_z the molar mass of the component in g mol⁻¹, n_z the number of electrons contained in component z, A the area of the electrode in square metre, J_app the applied current density in ampere per square metre, F the faraday constant of 96485 C mol⁻¹, [Gly_c;y] as the glycerol concentration in the cathode at a certain time point. This equation takes the occurrence and disappearance of 1,3-PDO in each reactor compartment into account. 1,3-PDO is an uncharged molecule and is able to diffuse along the concentration gradient over the ion exchange membranes. For glycerol, a similar consideration was made, but concentrations were negligible in the middle and anode compartment (~100 mg L⁻¹) and, thus, not taken into account in the calculations. The observed low diffusion of glycerol is attributed to fast consumption in the cathode and, thus, the concentration gradient was less strong as compared with the 1,3-PDO concentration gradient. CE was calculated in a similar fashion (i.e., considering an overall production from current and glycerol) for hydrogen production by taking the partial pressure and the gas flow rate into account. A separate efficiency for only current (CE_j) or glycerol (CE_gly) as electron donor can also be calculated by considering the other source as 0 in the denominator of Eq. 1.

In this study, the microbial inoculum used for bioelectrochemical synthesis of 3-carbon compounds had been through a selection and enrichment cycle. In a first cycle, the mixed-microbial community sample from a wastewater treatment plant of an oleochemical industry was cultured at 28°C and pH 6.0 in minimal medium with glycerol as an only carbon source for a period of 11 days. Mainly 1,3-PDO (0.5 g COD L⁻¹) and butyrate (0.3 g COD L⁻¹) were produced but glycerol was not entirely consumed (0.6 g COD L⁻¹) during this cycle (Transfer 0; Figure 2A). The headspace was dominated by nitrogen (54.2%) and hydrogen (17.8%), in this first cycle, and stayed constant throughout the whole enrichment procedure. Consumed glycerol mainly ended-up in microbial biomass production as dense bacterial growth (visual observation of turbidity) was observed after 11 days of incubation. Following a first transfer (Transfer 1; Figure 2A), microbial growth was still high, but within the same period of time (11 days), 1,3-PDO (9.0 g COD L⁻¹) as well as butyrate (7.7 g COD L⁻¹) concentrations were considerably higher. However glycerol (2.2 g COD L⁻¹) was still detected after 11 days. It was only after a second subsequent transfer (Transfer 2; Figure 2A) into fresh growth medium that glycerol was removed for 98% (0.3 g COD L⁻¹), which coincided with a high 1,3-PDO concentration (13.0 g COD L⁻¹) and a lower butyrate production (3.0 g COD L⁻¹) compared with the previous transfer. A third and last transfer into fresh growth medium did not improve glycerol degradation any further; a decrease in 1,3-PDO (8.6 g COD L⁻¹) and a slight increase in butyrate (3.7 g L⁻¹) production were detected (Transfer 3; Figure 2A). Additionally, CH₄ was never detected during the entire enrichment procedure.

Following this pre-enrichment of glycerol degraders, a similar enrichment procedure was performed using a larger volume of medium (40 mL) for 20 days. After 8 days of incubation, glycerol was mainly converted into 1,3-PDO (7.1 g COD L⁻¹) and butyrate (3.1 g COD L⁻¹). No further conversions occurred as the titers remained constant (Figure 2B).

The enriched mixed-microbial community was then transferred into modified medium from Dennis et al. (2013) to acclimate the community to the BES medium for a period of 12 days. After 7 days of incubation, a large fraction of glycerol (15.8 g COD L⁻¹) was consumed and 1,3-PDO production reached up to 8.2 g COD L⁻¹ and butyrate to 3.8 g COD L⁻¹ (Figure 2C).

In both reactors, headspace gas composition was very similar to each other (Figure 3). Hydrogen gas dominated the headspace of the reactors in the range of 60–88%, with the remainder being N₂ (10–38%) and CO₂ (0.1–4.3%). No methane was detected in the 2C and the 3C reactors, indicating successful inhibition of the methanogenic microbial community during the enrichment phase. The hydrogen production rate was 2.8 ± 0.6 L m⁻² day⁻¹ for 2C and 2.7 ± 0.7 L m⁻²day⁻¹ for 3C which corresponds to an overall CE for H₂ production of 16.7 ± 6.2% for 2C and 14.6 ± 4.6% for 3C (Eq. 1).

In both reactors, operated under the same conditions, 1,3-PDO reached similar and high titers, production rates, and CE (Eq. 1). After 13 days, 49.5 g COD L⁻¹ of 1,3-PDO (Figure 4C; produced at 33.2% CE and 380.5 g COD m⁻² day⁻¹; per unit projected surface area of electrode) with a yield of 0.72 mol_1,3-PDO mol⁻¹_glycerol was detected in the 2C bioelectrochemical reactor. During the same period, 50.8 g COD L⁻¹ 1,3-PDO (Figure 4A; produced at 34.1% CE and 391.0 g COD m⁻² day⁻¹) with a production yield of 0.71 mol_1,3-PDO mol⁻¹_glycerol was detected in the 3C bioelectrochemical reactor. A very low concentration of 1,3-PDO (0.1 g COD L⁻¹) and no glycerol were measured during this time in the extraction compartment medium of the 3C reactor, showing that the AEM is impervious to glycerol. From day 15, glycerol started to accumulate in the cathode compartment of both reactors and 1,3-PDO synthesis slowed down.

At day 13, oxidative products from glycerol electro-fermentation were dominated by propionate with 4.3 and 9.6 g COD L⁻¹ (6.2 g COD L⁻¹ in catholyte and 3.4 g COD L⁻¹ in extraction compartment; Figure 4B), respectively, in 2C and 3C reactors. These values represent 31.8 and 47.8%, respectively, of total volatile fatty acids (VFA) present in 2C and 3C reactors. Propionate production in both reactor configurations was rather high compared with what was previously obtained during the enrichment phase of this study. In the 2C reactor, butyrate concentration reached a maximum of 5.8 g COD L⁻¹ after 18 days of incubation, in accordance with the maximum titers previously obtained during the enrichment phase. During the same period, in the 3C reactor, the total titer of butyrate was much lower, 3.4 g COD L⁻¹ (1.7 g COD L⁻¹ in catholyte and 1.7 g COD L⁻¹ in extraction compartment; Figure 4B).

From day 11 to day 29 (end of experiment), a considerable increase of organic acids was detected in the extraction compartment with the main component being propionic acid, which increased in concentration from 3.4 to 13.1 g COD L⁻¹ in the extraction compartment of the 3C reactor (Figure 4B). After 29 days, propionate had been extracted at a rate of 162 g m⁻²h⁻¹ (Figure 4D). During this period, propionate was found to accumulate from 4.3 to 9.5 g COD L⁻¹ in the 2C reactor (Figure 4C), whereas its concentration was constant around 4.8 ± 0.4 g COD L⁻¹ (n = 6) in catholyte of the 3C reactor (Figure 4A). The concentration of total VFA (lactate, formate, acetate, propionate, butyrate) increased from 11.4 to a maximum of 26.6 g COD L⁻¹ in catholyte of the 2C reactor, while it stabilized to 12.7 ± 0.8 g COD L⁻¹ (n = 6) in catholyte of the 3C reactor. At day 29, 9.5 and 17.7 g COD L⁻¹ of propionate (from which 13.4 g COD L⁻¹ was measured in extraction medium) was measured in the 2C and 3C reactor, respectively. After 29 days, based on VFA considered/detected in the study, the total VFA was extracted with an efficiency of 34.2%. Throughout this period, the concentration of 1,3-PDO remained constant in the 3C reactor, from 51.3 (mainly present in catholyte) to 52.5 g COD L⁻¹ (47.4 g COD L⁻¹ in catholyte and 5.1 g COD L⁻¹ in extraction compartment), representing 58.7 ± 0.1% (n = 6) of total COD in the catholyte and middle compartment. In contrast, the 1,3-PDO concentration dropped from 49.5 g COD L⁻¹ (77.7% of total COD) to 38.5 g COD L⁻¹ (34.2% of total COD) in the 2C reactor.

From day 18 until the end of the experiment, glycerol started to accumulate in the catholyte of both systems. Nevertheless, 25.7 and 46.4 g COD L⁻¹ of glycerol was still consumed or degraded in the 2C and 3C reactor, respectively.