Comparative assessment of two biotrickling filters for siloxanes removal: Effect of the addition of an organic phase

1 Comparative assessment of two biotrickling filters for siloxanes 1 removal: effect of the addition of an organic phase 2 Celia Pascuala,b, Sara Canterac, Raúl Muñoza, b, Raquel Lebreroa, b* 3 a Department of Chemical Engineering and Environmental Technology, University of Valladolid, Dr. 4 Mergelina s/n., Valladolid 47011, Spain. 5 b Institute of sustainable processes, University of Valladolid, Dr. Mergelina s/n., Valladolid 47011, Spain 6 c Laboratory of Microbiology, Wageningen University and Research Center, The Netherlands7 *Author for correspondence: [email protected] 9 Keywords: Biogas upgrading, Biotrickling filter, Silicone oil, Siloxanes, Two-phase 10 partitioning bioreactor. 11 12 Abstract 13 Biogas produced at wastewater treatment plants and landfills contains trace levels of volatile 14 methyl siloxanes (VMS) that are responsible for abrasion, corrosion and erosion of equipment 15 during biogas storage and combustion. This research comparatively evaluated the removal of the 16 most common VMS (L2, L3, D4, and D5) under aerobic conditions in a conventional biotrickling 17 filter (BTF) and a two-phase partitioning BTF (TP-BTF) with silicone oil (at 30 %) as organic 18 phase. The TP-BTF showed a superior performance compared to the conventional BTF, 19 increasing the total VMS removal from < 30 % in the BTF up to ~70 % in the TP-BTF. The 20 highest REs in the TP-BTF were recorded for D4 and D5, reaching values of 80-90 %, 21 corresponding to ECs between 0.12 and 0.17 g.m-3.h-1. Slightly lower values were obtained for 22 L3 (70-80 %), and the lowest performance was recorded for L2 (20-60%) due to the high vapor 23 pressure of this siloxane and therefore its lower affinity by the organic phase. Surprisingly, despite 24 the different inocula used, a similar microbial community was found by the end of operation of 25 both BTFs, with KMBC-112, Reynarella and Chitinophaga as the dominant genera. 26 27 28 29 © 2020 Elsevier. This manuscript version is made available under the CC-BY-NC-ND 4.0 2 1. Introduction 30 Biogas produced at wastewater treatment plants (WWTPs) and landfills contains trace 31 amounts of undesirable contaminants such as hydrogen sulfide, NH3, halogenated 32 hydrocarbons, volatile organic compounds and siloxanes (Muñoz et al., 2015; 33 Ryckebosch et al., 2011). Of them, siloxanes lead to severe detrimental effects when using 34 biogas as an energy vector in turbines, microturbines or fuel cells (Ajhar et al., 2010). 35 Polydimethylsiloxanes (PDMS) are organosilicon compounds widely used in household 36 and industrial products, such us cosmetics and personal care products, detergents, 37 building materials and textiles (Li et al., 2014; Soreanu et al., 2011). When these products 38 reach WWTPs, organosilicon hydrolyzation results in the formation of volatile methyl 39 siloxanes (VMS), a group of low molecular weight compounds that are eventually 40 volatilized to biogas (Soreanu et al., 2011). These VMS are further oxidized to crystalline 41 deposits of silicon dioxide (SiO2) during biogas combustion (Eq. 1), which cause 42 abrasion, corrosion and erosion of turbine parts. Furthermore, SiO2 can deactivate 43 catalytic converts resulting in undesirable gas emissions of CO and SOx (Läntelä et al., 44 2012; Li et al., 2014). 45 ((CH3)2SiO)n + 4nO2  n SiO2 + 2nCO2 + 3nH2O 46 Eq. 1. VMS oxidation reaction during combustion process, where n indicates the 47 numbers of Si atoms (n = 3 –6) (Ruiling et al., 2017). 48 49 The main VMS present in biogas are hexamethyldisiloxane (L2), octamethyltrisiloxane 50 (L3), hexamethylcyclotrisiloxane (D3), octamethylcyclotetrasiloxane (D4) and 51 decamethylcyclopentasiloxane (D5), D4 and D5 being the most abundant in biogas from 52 landfills and anaerobic digesters (Shen et al., 2018). VMS concentration in raw biogas 53 typically ranges from 16 up to 400 mg.m-3 (Dewil et al., 2006). These values far exceed 54 3 the maximum VMS concentration allowed for biomethane injection into natural gas grid, 55 which according to the most recent European standard 16723 must be lower than 1 mg 56 Si.m-3 (Standardization, 2016). 57 Therefore, the removal of siloxanes prior biogas usage is of utmost importance in any 58 biogas-to-energy application, and mandatory when the upgraded biogas is envisaged as 59 vehicle fuel or as a substitute of natural gas. In this context, the main VMS abatement 60 technologies commercially available are based on physical-chemical processes, such as 61 adsorption, absorption and cryogenic condensation. Despite the successful removal 62 efficiencies achieved, these technologies only transfer the undesired compounds from one 63 phase to another, resulting in hazardous wastes that require further treatment, which 64 entails high investment and operating costs (Gaj, 2017; Muñoz et al., 2015). 65 On the contrary, biotechnologies have arisen as a cost-effective and environmentally 66 friendly alternative to these physical-chemical processes. Whereas little is known on the 67 microbiology underlying the biodegradation of VMS, there are several studies focused 68 on the implementation of biotechnologies for continuous VMS removal. For instance, 69 Accettola et al (2008) and Popat and Deshusses (2008) operated aerobic biotrickling 70 filters (BTF) inoculated with isolated D4-degrading bacteria, reaching removals 71 efficiencies of this VMS of 20-43 % at empty bed residence times (EBRTs) ranging from 72 2.2 up to 19 min. Li et al (2014) significantly improved these preliminary results 73 achieving REs over 74 % for D4 at an EBRT of 13.2 min in an aerobic BTF inoculated 74 with Pseudomonas aeruginosa S240. This superior process performance was attributed 75 to the presence of rhamnolipids, biosurfactants produced by P. aeruginosa that could have 76 increased the mass transfer of D4 from the gas phase to the aqueous phase. Overall, these 77 investigations suggested that the main bottleneck during biological VMS removal is the 78 low solubility of these compounds in the aqueous phase, and hence their poor availability 79 4 to the microbial community. Since bioreactors operation at high EBRTs results in 80 prohibitive reactor volumes, the addition of a non-aqueous (organic) phase with a high 81 affinity for the VMS represents as a potential solution to overcome mass transfer 82 limitation. In this context, the superior performance of two-phase partitioning bioreactors 83 (TPPB) for the removal of hydrophobic volatile organic compounds such as hexane has 84 been consistently demonstrated during the past decade. TPPBs are based on boosting the 85 mass transfer of the target gas compound by adding an organic water-immiscible and non86 volatile phase with a high affinity of the gas pollutant (Muñoz et al., 2007). Organic 87 solvents such as hexadecane and silicone oil have been employed as organic phase in 88 TPPBs. A recent research demonstrated the enhancement of chlorobenzene 89 biodegradation by using a water-silicone oil biphasic system (Ye et al., 2019). 90 Nevertheless, TPPBs have not been yet applied for the removal of VMS from biogas. 91 This study aimed at comparatively evaluating the removal of a mixture of VMS (L2, L3, 92 D4, and D5) by using both a conventional BTF (without organic phase) and a two-phase 93 BTF (TP-BTF). The organic phase selected in the TP-BTF was silicone oil, due to its high 94 affinity for VMS. In addition, the bacterial community structure of both bioreactors was 95 analyzed with the aim of identifying the most representative microorganisms involved in 96 siloxanes degradation. This research constitutes an important contribution to the field of 97 biogas upgrading since it explores and demonstrates the viability of siloxanes 98 biodegradation in two-phase BTFs, a promising, sustainable and innovative technology 99 scarcely studied so far 100 101 102 103 2. Materials and Methods 104 5 2.1. Mineral Salt Medium 105 The mineral salt medium (MSM) was composed of (g.L-1): KH2PO4, 0.7; K2HPO4·3H2O, 106 0.917; KNO3, 3; NaCl, 0.2; MgSO4·7H2O, 0.345; CaCl2·2H2O, 0.026; and 2 mL.L-1 of a 107 micronutrients solution containing (g.L-1): EDTA, 0.5; FeSO4·7H2O, 0.2; ZnSO4·7H2O, 108 0.01; MnCl2·4H2O, 0.003; H3BO3, 0.003; CoCl2·6H2O, 0.02; CuCl2·2H2O, 0.001; NiCl2 109 ·6H2O, 0.002; NaMoO4·2H2O, 0.003. All the chemicals used for the preparation of the 110 MSM were purchased from Panreac (Barcelona, Spain). L2 (98.5% purity), L3 (98 % 111 purity), D4 (98 % purity) and D5 (97 % purity) were obtained from Sigma Aldrich (San 112 Luis, EEUU). 113 114 2.2 Culture enrichment conditions 115 Three aerobic batch tests were operated to enrich a VMS degrading culture using 116 activated sludge as inoculum (Valladolid WWTP, Spain). For this purpose, 1.2 L bottles 117 were filled with 0.2 L of mineral salt medium at a pH  7 and inoculated with 100 mL of 118 activated sludge. A 20 mL gas mixture of VMS (containing trimethylsilanol (TMSOL), 119 L2, L3, L4, L5, D4 and D5 at individual concentrations of ~50 mg.m-3) was added to the 120 headspace as the only carbon and energy source at an initial total concentration of ~350 121 mg VMS.m-3. The enrichment was conducted for 360 days. Both gas and liquid phases 122 were replaced by day 39, 174 and 230. For this purpose, the cultivation broth was 123 centrifuged for 10 min at 10000 rpm and the pellet was resuspended in 0.3 L of fresh 124 mineral medium. The bottles headspace was flushed with fresh air prior supplementation 125 of 20 mL of VMS gas mixture. By days 95 and 304, 20 mL of VMS gas mixture were 126 added to the headspace without replacing neither the headspace nor the liquid phase. 127 VMS, CO2, O2 and N2 concentrations were periodically analyzed in the headspace of the 128 6 bottles, and samples of the cultivation broth were periodically withdrawn to determine 129 pH, NO3-, NO2-, TN, TOC and IC concentrations. 130 131 2.3. Experimental setup and operating procedure. 132 The experimental systems (Fig. 1) consisted of a cylindrical PVC column of 2 L of 133 working volume (8.4 cm diameter, 37.5 cm height). A 1.2 L holding tank magnetically 134 stirred at 100 rpm was used as MSM reservoir. The MSM in the BTF, and the mixture 135 MSM + silicone oil in the TP-BTF, were continuously recycled to the top of the column 136 by a peristaltic pump at a linear velocity of 2 m.h−1. The VMS-loaded inlet air stream was 137 prepared by injecting a liquid mixture containing L2, L3, D4 and D5 with a syringe pump 138 (Fusion 100, Chemyx Inc. USA) into a 33 mL.min-1 air stream controlled by means of a 139 rotameter. The VMS-loaded stream entered a mixing chamber and was subsequently fed 140 at the bottom of the column countercurrently with the trickling liquid flow. 141 7 142 143 Fig. 1. Schematic representation of the experimental set-up. (1) Biotrickling filter, (2) 144 nutrient reservoir, (3) syringe pump, (4) rotameter, (5) mixing chamber, (6) air 145 compressor, (7) peristaltic pump, (8) gas sampling ports. 146 147 An abiotic test was initially performed to discard the possibility of siloxanes removal by 148 photolysis or adsorption, and to ensure that there was no biological activity in the system 149 prior to inoculation. For this purpose, the test was initiated with the empty PVC column, 150 while sterile packing material and MSM were added in subsequent steps. 151 The BTF was inoculated with the previously enriched siloxane-degrading culture. The 152 system was operated during 160 days in five different stages (Table 1) at an EBRT of 1 h 153 and a periodical replacement of the culture broth with fresh MSM of 200 mL every three 154 days (equivalent to a hydraulic retention time of 18 days). During stages S1, S3 and S5, 155 Air Gas outlet Liquid recycling 1 2 3 5 4 6 7 8 8 8 the system was fed with a VMS-loaded stream with a total VMS concentration between 156 500 and 700 mg.m-3. By day 81, the original enriched culture was used to re-inoculate the 157 system. During stage S2, VMS feeding was stopped in order to evaluate potential 158 emissions of VMS adsorbed in the packing material or biofilm, and CO2 production. 159 During this period, no MSM exchange was performed. Finally, a two-fold increase in the 160 VMS concentration up to ~1300 mg.m-3 was implemented during stage S4. 161 Table. 1. Experimental conditions tested in BTF. 162 Feed Stream Time course (days) VMS Concentration (mg.m-3) S1 VMS loaded air stream 0 - 46 515 ± 126 S2 VMS free air stream 47 - 76 - S3 VMS loaded air stream 77 - 124 719 ± 203 S4 VMS loaded air stream 125 -137 1288 ± 217 S5 VMS loaded air stream 138 - 160 651 ± 127 163 In the particular case of the TP-BTF, constructed with 30 % of silicone oil 20 cts, 164 siloxanes removal and CO2 production were observed by day 10. Therefore, no additional 165 inoculum was added and the TP-BTF was operated in three different stages at an EBRT 166 of 1 h (Table 2). From day 84 onwards, 200 mL of the culture broth were withdrawn 167 every three days (equivalent to a hydraulic retention time of 18 days). The silicone oil 168 was recovered by settling, supplemented with fresh MSM up to 200 mL and returned to 169 the system. Thus, the silicone oil initially added to the TP-BTF was reused during the 170 entire experiment without necessity of replacement or further supply. During stages S1 171 and S3, the system was fed with a VMS-loaded stream at a concentration of ~650 mg.m172 3, while during S2 both VMS feeding and culture broth replacement were stopped. 173 174 Table. 2. Experimental conditions in TP-BTF. 175 Feed Stream Time course VMS Concentration 9 (days) (mg.m-3) S1 VMS loaded air stream 0 - 21 651 ± 181 S2 Clean air stream 22 - 70 - S3 VMS loaded air stream 70 - 127 625 ± 137 176 Inlet and outlet VMS and CO2 gas concentrations were daily analyzed. The samples 177 were taken directly from the VMS-loaded air stream by means of a gas sampling port 178 located at both inlet and outlet gas streams of the BTFs The TOC, TN, nitrite and nitrate 179 concentrations and the pH of the cultivation broth were periodically analyzed every 180 three days. 181 182 2.4. Analytical procedure 183 VMS gas concentration was analyzed in a Bruker 3900 gas chromatograph (Palo Alto, 184 USA) equipped with a flame ionization detector and a HP-5-MS (30 m × 0.25 mm × 0.25 185 µm) column. Both the detector and injector temperatures were maintained constant at 250 186 ºC. The oven temperature was initially set at 40 ºC for 2.0 min, then increased at 20 ºC 187 min-1 up to 180 ºC, maintained for 1 min and increased again at 20 ºC min-1 up to 200 ºC. 188 Finally, this temperature was maintained for 0.5 min. N2 was used as the carrier gas at a 189 flow rate of 1 mL.min-1. CO2 and O2 gas concentrations were determined in a Bruker 430 190 gas chromatograph (Palo Alto, USA) coupled with a thermal conductivity detector and 191 equipped with a CP-Molsieve 5A (15 m × 0.53 mm × 15 µm) and a P-PoraBOND Q (25 192 m × 0.53 mm × 10 µm) columns. Oven, detector and injector temperatures were 193 maintained constant at 45, 200 and 150 °C for 5 min, respectively. Helium was used as 194 the carrier gas at a flow of 13.7 mL.min-1. 195 The pH of the cultivation broth was analyzed using a glass membrane electrode PH 196 BASIC 20 (Crison, Barcelona, Spain). TOC, IC and TN concentrations were measured 197 16 emission was observed during S2, probably due to the desorption of the VMS from the 303 silicone oil (Fig. 4). 304 305 306 Fig. 4. Time course of L2, L3, D4 and D5 inlet (●) and outlet (○) concentration in the 307 TP-BTF. Vertical lines represent standard deviation from triplicate measurements. 308 309 0 50 100 150 200 250 300 350 400 0 20 40 60 80 100 120 Concentration (mg.m-3) L2 S1 S2 S3 0 50 100 150 200 250 300 350 400 0 20 40 60 80 100 120 L3 S1 S2 S3 0 50 100 150 200 250 300 350 400 020 40 60 80 100 120 Concentration (mg.m-3) Time (d) D4 S1 S2 S3 0 50 100 150 200 250 300 350 400 0 20 40 60 80 100 120 Time (d) D5 S1 S2 S3 17 310 Fig. 5. Time course of total VMS inlet (●) and outlet (○) concentration in the TP-BTF. 311 Vertical lines represent standard deviation from triplicate measurements. 312 313 The biodegradation of VMS resulted in a CO2 production of 2.24 ± 0.29 and 2.26 ± 0.96 314 g.m-3.h-1 during S1 and S3, respectively. Similarly, pH remained constant at 7.1 ± 0.3 315 during the entire experiment. TN concentration in the cultivation broth decreased from 316 402.7 to 214.3 mg.L-1 by day 84. Afterwards, a gradual increase was recorded up to a 317 maximum concentration of 376.9 mg.L-1 by day 127 due to the mineral medium 318 exchange. Similarly, N-NO3¯ concentrations of 403.3, 229.8 and 343.2 mg.L-1 were 319 recorded by days 0, 84 and 127, respectively. No NO2production was observed. 320 Finally, TOC concentration in the cultivation broth increased initially from 4.3 to 219.5 321 mg.L-1 by day 84, gradually decreasing afterwards to 100.7 mg.L-1 by day 127. IC 322 concentration remained constant at 6.5 ± 4.2 mg.L-1 throughout the entire experiment. 323 324 3.4 Analysis of the microbial community 325 0 200 400 600 800 1000 1200 1400 020 40 60 80 100 120 Concentration (mg.m-3) Time (d) S1 S2 S3 18 Activated sludge (AS) was used as inoculum for the enrichment of a siloxane degrading 326 consortium. The dominant genera in the AS were Hydrogenophaga, Comamonas, 327 Albidiferax and Chitinophaga, with abundances of 27.7, 15.9, 7.8 and 2.3 %, respectively 328 (Fig.6). However, during the enrichment of BTF-InA (day 278) and BTF-InB (day 360) 329 a significant shift in the microbial population was observed. In these samples, the genera 330 Hydrogenophaga, Comamonas and Albidiferax represented less than 1 % of the total 331 population, while bacteria from the genus Chitinophaga increased their abundance up to 332 11.8 % in BTF-InA and to 4.7 % in BTF-InB. Regardless of the enrichment duration, both 333 samples presented a similar population structure, which consisted mainly of 5 genera 334 (data shown as relative abundance in BTF-InA and BTF-InB, respectively): 335 Sphingomonas (17.1 and 29.9 %), Achromobacter (10.5 and 14.4 %), Bacteroides (16.4 336 and 3 %) and Mucilaginibacter (4.4 and 6.3 %), as well as the above mentioned 337 Chitinophaga (Fig. 5). Interestingly, by the end of the experimental period (stage S5), 338 the diversity and richness of the bacterial population in BTF increased in terms of absolute 339 abundance. The continuous exposure to siloxanes supported the growth of genera with 340 negligible abundances in the inocula, such as Reyranella (10.1 %), Solimonas (6.8 %), 341 Ferruginibacter (5.6 %), Mycobacterium (5.0 %) and an uncultured genus from the 342 family Acidithiobacillaceae, KCMB-112 (4.9 %) (SILVA database Accession Nr: 343 FJ914601). Moreover, the genus Chitinophaga (5.1 %) still represented an important 344 share of the population in the sample BTF-END. 345 Interestingly, biomass grew in the TP-BTF without previous inoculation using siloxanes 346 as the only carbon and energy source. The main genera retrieved in stage S1 (TP-S1) were 347 Pseudoxanthomonas (21.2 %), Proteobacter (9.9 %), Rhizobium (9.4 %) and the genus 348 Chitinophaga (5.9 %). Nevertheless, after operation of the TP-BTF (TP-END) the 349 bacterial population shifted towards a more specialized community that was similar to the 350 19 bacterial population independently reached during the operation of BTF. The main genera 351 were KCMB-112 (20 %), Flavobacterium (11.2 %) Reyranella (7.0 %) and Chitinophaga 352 (6.9 %). 353 354 355 Fig. 6. Heatmap showing the differential number of sequences of the most significant 356 22 bacterial genera in the activated sludge (AS), the enriched culture used as inoculum 357 of the BTF (BTF-InA, BTF-InB), the end of operation of the BTF (BTF-END) and the 358 TP-BTF operation (TP-S1, TP-END). OTUs with absolute abundances < 1 were not 359 included in the data analysis. 360 361 4. Discussion 362 A clear improvement in the siloxanes abatement performance of the BTF was observed 363 with the addition of silicone oil at 30 %. In this sense, the presence of the organic phase 364 resulted in an increase in the total VMS RE from 20 % in the BTF to 70% in the TP-BTF, 365 which resulted in an EC 5 fold higher. The re-inoculation of the BTF by day 81 had no 366 effect on performance of the system, the poor VMS REs recorded being associated to the 367 20 low solubility of the VMS in the aqueous phase and hence to their reduced mass transfer. 368 Several studies have confirmed that the presence of an organic phase in a TPPB increases 369 the elimination performance of poorly soluble contaminants such as CH4, toluene and 370 styrene (Cantera et al., 2016; San-Valero et al., 2017; Nourmohammadi et al., 2018). To 371 the best of our knowledge, this is the first study validating the potential of a silicon oil372 based BTF for VMS removal. Moreover, the negligible VMS desorption observed during 373 S2 (no VMS feeding) supported the high affinity of silicone oil for siloxanes. This result, 374 along with the significantly higher CO2 production recorded in the TP-BTF compared to 375 the BTF, confirmed biodegradation as the main mechanism of VMS removal. In addition, 376 the higher TOC concentration in the trickling solution of the TP-BTF (~ 150 mg.L-1 vs. ~ 377 4 mg.L-1 in BTF) further evidenced the biological activity. Based on the results obtained 378 by Accettola et al (2008) and Li et al (2014), the increase in TOC concentration was 379 attributed to Si-containing metabolites such as silicic acid. At this point is should be 380 stressed that the aqueous solubility of silicone oil is negligible. 381 When analyzing the individual removal performance of the different VMS, the lowest 382 enhancement was observed for L2, increasing from an overall RE of ~10 % in BTF to 27 383 % in TP-BTF. This improvement in RE was significantly higher for L3, reaching an 384 overall RE of ~ 76 % in the TP-BTF compared to the average RE of 12 % recorded in the 385 BTF. Nevertheless, the most remarkable enhancement was obtained for D4 and D5, with 386 REs between 85 and 90 % in the TP-BTF vs 13 and 18 % in the BTF, respectively. These 387 cyclic siloxanes are typically the most abundant VMS in biogas, and therefore the 388 biological removal of lineal VMS has not been reported to date. For instance, Popat and 389 Deshusses (2008) obtained D4 removals of 43 % at an EBRT of 19.5 min under aerobic 390 conditions, while Li et al (2014) significantly improved previous results reaching D4 REs 391 > 74 % in an aerobic BTF operating at an EBRT of 13.2 min. This enhanced performance 392 21 was attributed to the presence of rhamnolipids in the trickling solution, which likely 393 fostered the mass transfer of D4 from the gas to the aqueous phase. The effect of the gas 394 residence time and the packing material in the removal of both D4 and D5 was studied 395 by Santos-Clotas et al (2019) in an anoxic BTF using nitrate as electron acceptor. The 396 highest REs obtained were 13 % and 37 % for D4 and D5, respectively, at the highest 397 EBRT (14.5 min). Moreover, the addition of activated carbon to the packing material 398 resulted in an increased mass transfer of D4 and D5, which supported REs of 16 % and 399 45 %, respectively. The low L2 removal was attributed to its high vapor pressure, which 400 hindered the solubility of this compound in the organic phase (Table 3). As demonstrated 401 by Rojas Devia and Subrenat (2013), the lower the vapor pressure of siloxanes the easier 402 their removal from the gas phase by absorption into different oils (L2>L3>D4>D5). 403 These researchers also reported a significant effect of the temperature on VMS mass 404 transfer, with a significant increase in absorption efficiency when operating at the lowest 405 temperature, which provided removals of 80 % and 60 % for D4 and L2, respectively. 406 407 Table 3. Physical properties of volatile methyl siloxanes present in biogas. 408 Compound Formula Boiling point (ºC)a Molar Mass (g.mol-1) Saturated vapor pressure at 25°C (Pa)a Water solubility at 25 ºC (mg.L-1)a Hexamethyldisiloxane (L2) C6H18OSi2 106.9 162.4 5626.2 0.93 Octamethyltrisiloxane (L3) C8H24O2Si3 153.0 236.5 445.0 0.034 Decamethyltetrasiloxane (L4) C10H30O3Si4 194.0 310.7 50.0 0.00674 Dodecamethylpentasiloxane (L5) C12H36O4Si5 232.0 384.8 9.0 0.000309 Hexamethylcyclotrisiloxane (D3) C6H18O3Si3 135.2 222.5 471.0 1.56 Octamethylcyclotetrasiloxane (D4) C8H24O4Si4 175.7 296.6 132.0 0.056 Decamethylcyclopentasiloxane (D5) C10H30O5Si5 211.2 370.8 23.2 0.017 Dodecamethylcyclohexasiloxane (D6) C12H36O6Si6 245.1 444.9 4.0 0.005 a Adapted from Ruiling et al (2017). 409 410 22 Finally, the microbial analysis showed significant differences between the bacterial 411 community of the enriched inocula and those retrieved by the end of operation of the 412 bioreactors. This could be attributed to the different VMS mixture employed: while the 413 enriched culture was fed with a wide range of VMS (TMSOL, L2, L3, L4, L5, D4 and 414 D5), only L2, L3, D4 and D5 were fed to the BTF and TP-BTF. In this sense, the 415 enrichment test promoted the growth of the members from the genus Sphingomonas and 416 Chitinophaga in BTF-InA and BTF-InB, which represented 40 % of the total population. 417 In the case of BTF and TP-BTF, the continuous exposition to siloxanes shifted the initial 418 microbial population to a similar bacterial community regardless of the inoculum used. 419 KMBC-112, Reyranella and Chitinophaga were the main genera favored by VMS 420 exposure (L2, L3, D4 and D5), representing 30 % of the population. Although no previous 421 studies have pointed out any of the genera found in this study as VMS degraders, the 422 genus Chitinophaga, Sphingomonas and Reyranella consist of highly versatile 423 microorganisms capable of growing in a wide range of environments and have been found 424 before in bioreactors devoted to the removal of volatile organic compounds such as 425 toluene or dichloromethane (Cheng et al., 2018; Xu et al., 2019). In the case of KMBC426 112, it has been retrieved from urban deposits and contaminated soils (Marti et al., 2017). 427 In this regard, siloxanes can be efficiently eliminated by a bespoke consortium that will 428 vary based on the target VMS. 429 430 5. Conclusions 431 This study demonstrated the superior siloxanes abatement performance of a two-phase 432 partitioning BTF compared to a conventional BTF. While the BTF achieved a total VMS 433 removal lower than 30 %, this value increased up to ~ 70 % due to the addition of a non434 aqueous phase (i.e. silicon oil), corresponding to an EC 5× higher than that of the BTF. 435 23 The highest REs were recorded for D4 and D5, reaching values between 80-90 % 436 throughout the entire experiment (corresponding to ECs between 0.12-0.17 g.m-3.h-1). 437 The RE of L3 was slightly lower (70-80 %), while only 20-60 % of L2 was removed 438 depending on the operating conditions. This outstanding performance of the TP-BTF was 439 associated to the presence of silicone oil that boosted the mass transfer of VMS from the 440 gas phase to the liquid phase. The removal of L2 was hindered by its higher vapor pressure 441 compared to L3, D4 and D5, decreasing the solubility of this compound in the organic 442 phase. Finally, bacterial genera KMBC-112, Reynarella and Chitinophaga represented 443 more than 30 % of the total population retrieved by the end of operation of both BTF 444 despite the different inocula and operating conditions in both bioreactors. Up to date, no 445 evidence of siloxanes biodegradation capacity of these genera has been reported. 446 447 Acknowledgements 448 This work was supported by the regional government of Castilla y León and the EU449 FEDER programme and by the URBIOFIN project. The project has received funding 450 from the Bio Based Industries Joint Undertaking under the European Union’s Horizon 451 2020 research and innovation program under grant agreement No 745785. The authors 452 also thank Ramon Areces Foundation for the financial support of this work via sponsoring 453 of Sara Cantera Postdoctoral grant. 454 455 456 References 457 Accettola, F., Guebitz, G.M., Schoeftner, R., 2008. Siloxane removal from biogas by 458 24 biofiltration: Biodegradation studies. Clean Technol. Environ. Policy 10, 211–218. 459 doi:10.1007/s10098-007-0141-4 460 Ajhar, M., Travesset, M., Yüce, S., Melin, T., 2010. Siloxane removal from landfill and 461 digester gas - A technology overview. Bioresour. Technol. 101, 2913–2923. 462 doi:10.1016/j.biortech.2009.12.018 463 Cantera, S., Estrada, J.M., Lebrero, R., García-Encina, P. a., Muñoz, R., 2016. 464 Comparative performance evaluation of conventional and two-phase hydrophobic 465 stirred tank reactors for methane abatement: Mass transfer and biological 466 considerations. Biotechnol. Bioeng. 113, 1203–1212. doi:10.1002/bit.25897 467 Cheng, C., Zhou, Z., Pang, H., Zheng, Y., Chen, L., Jiang, L.M., Zhao, X., 2018. 468 Correlation of microbial community structure with pollutants removal, sludge 469 reduction and sludge characteristics in micro-aerobic side-stream reactor coupled 470 membrane bioreactors under different hydraulic retention times. Bioresour. 471 Technol. 260, 177–185. doi:10.1016/j.biortech.2018.03.088 472 Cole, J.R., Wang, Q., Cardenas, E., Fish, J., Chai, B., Farris, R.J., Kulam-Syed473 Mohideen, A.S., McGarrell, D.M., Marsh, T., Garrity, G.M., Tiedje, J.M., 2009. 474 The Ribosomal Database Project: Improved alignments and new tools for rRNA 475 analysis. Nucleic Acids Res. doi:10.1093/nar/gkn879 476 Dewil, R., Appels, L., Baeyens, J., 2006. Energy use of biogas hampered by the 477 presence of siloxanes. Energy Convers. Manag. 47, 1711–1722. 478 doi:10.1016/j.enconman.2005.10.016 479 Gaj, K., 2017. Applicability of selected methods and sorbents to simultaneous removal 480 of siloxanes and other impurities from biogas. Clean Technol. Environ. Policy 19, 481 2181–2189. doi:10.1007/s10098-017-1422-1 482 Läntelä, J., Rasi, S., Lehtinen, J., Rintala, J., 2012. Landfill gas upgrading with pilot483 25 scale water scrubber: Performance assessment with absorption water recycling. 484 Appl. Energy 92, 307–314. doi:10.1016/j.apenergy.2011.10.011 485 Li, Y., Zhang, W., Xu, J., 2014. Siloxanes removal from biogas by a lab-scale 486 biotrickling filter inoculated with Pseudomonas aeruginosa S240. J. Hazard. Mater. 487 275, 175–184. doi:10.1016/j.jhazmat.2014.05.008 488 Marti, R., Bécouze-Lareure, C., Ribun, S., Marjolet, L., Bernardin Souibgui, C., Aubin, 489 J.B., Lipeme Kouyi, G., Wiest, L., Blaha, D., Cournoyer, B., 2017. Bacteriome 490 genetic structures of urban deposits are indicative of their origin and impacted by 491 chemical pollutants. Sci. Rep. 7, 1–14. doi:10.1038/s41598-017-13594-8 492 Muñoz, R., Meier, L., Diaz, I., Jeison, D., 2015. A review on the state-of-the-art of 493 physical/chemical and biological technologies for biogas upgrading. Rev. Environ. 494 Sci. Biotechnol. 14, 727–759. doi:10.1007/s11157-015-9379-1 495 Muñoz, R., Souza, T.S.O., Glittmann, L., Pérez, R., Quijano, G., 2013. Biological 496 anoxic treatment of O2-free VOC emissions from the petrochemical industry: A 497 proof of concept study. J. Hazard. Mater. 260, 442–450. 498 doi:10.1016/j.jhazmat.2013.05.051 499 Muñoz, R., Villaverde, S., Guieysse, B., Revah, S., 2007. Two-phase partitioning 500 bioreactors for treatment of volatile organic compounds. Biotechnol. Adv. 25, 501 410–422. doi:10.1016/j.biotechadv.2007.03.005 502 Pérez, R., Cantera, S., Bordel, S., García-Encina, P.A., Muñoz, R., 2019. The effect of 503 temperature during culture enrichment on methanotrophic polyhydroxyalkanoate 504 production. Int. Biodeterior. Biodegrad. doi:10.1016/j.ibiod.2019.04.004 505 Phandanouvong-Lozano, V., Sun, W., Sanders, J.M., Hay, A.G., 2018. Biochar does not 506 attenuate triclosan’s impact on soil bacterial communities. Chemosphere. 507 doi:10.1016/j.chemosphere.2018.08.132 508