Enhancement of swine manure anaerobic digestion using membrane-based NH 3 extraction

concentrations mediated by membrane operation induced an increase in CH 4 yield from 380.4 ± 84.9 up to 566.1 ± 7.8 NmLCH 4 g VS fed (cid:0) 1 . Likewise, COD and VS removal efficiencies significantly increased from 33.0 % ± 2.0 % and 25.7 % ± 2.3 % up to 61.8 % ± 1.3 % and 37.9 % ± 1.8 %, respectively. Interestingly, the decrease in NH 3 concentration entailed a complete assimilation of VFA.

• VFAs were completely removed from the anaerobic broth during NH 3 extraction.

Introduction
Today, there is an urgent need to substitute fossil fuels by renewable energy vectors in order to mitigate climate change and to protect the environment (Kang et al., 2020).In this context, while the main source of bioenergy in Earth is lignocellulosic biomass (Guldhe et al., 2017), biogas from the anaerobic digestion of animal manure, industrial or urban solid waste, sewage sludge, energy crops or wastewaters, etc., has recently attracted an increasing attention due to its versatility and high energy content (Kunatsa and Xia, 2022;Ye et al., 2021).Anaerobic digestion is based on the microbial hydrolysis and bioconversion of biodegradable organic matter in the absence of oxygen into biogas, which is primarily composed of methane and carbon dioxide (Reyes et al., 2015).Biogas is a renewable gas energy vector used to generate heat and electricity, which can replace (prior upgrading) natural gas in the gas networks and fossil fuels in transportation (Andriani et al., 2014).Biogas composition is determined by the oxidation-reduction state of the organic matter in the biodegradable substrate and the type of anaerobic digestion (AD) process (Jonsson et al., 1997).The hydrolysis of organic matter and the further ammonification of organic nitrogen generate high ammonia concentrations in the digestates, which entail a detrimental impact on water bodies (eutrophication and nitrate pollution) and air quality (NH 3 emissions) (Yenigün and Demirel, 2013).Indeed, NH 3 emissions can cause cancer, pulmonary diseases and are considered atmospheric precursors of N 2 O, which is a powerful greenhouse gas (Temkin et al., 2019).Indeed, the European Directive 2016/2284 will enforce reductions in NH 3 emissions up to 3 % between 2020 and 2029 and up to 16 % from 2030 onwards (European Parliament and Council, 2016).In addition, concentrations higher than 0.4 g NH 3 -N L − 1 can inhibit methane production during anaerobic digestion, which represents a technical limitation during the anaerobic digestion of livestock manure (Procházka et al., 2012;Sung and Liu, 2003).Methanogens are the most sensitive microorganisms exposed to NH 3 toxicity (Chen et al., 2008).For instance, total ammoniacal nitrogen concentrations of 1.7 -14 g L − 1 cause a 50 % reduction in methane production during anaerobic digestion (Angelidaki and Ahring, 1994;Chen et al., 2008;Hendriksen and Ahring, 1991;Sung and Liu, 2003).
Nowadays, multiple technologies are applied in order to reduce NH 3 concentrations from wastewaters.The implementation of NH 3 stripping or denitrification-nitrification processes entail high operational costs (Licon Bernal et al., 2016) and a severe impact to the environment.Therefore, the development of cost-efficient and sustainable technologies for NH 3 recovery from wastewater is mandatory in the current quest for creation of circular bioeconomy concepts.In this context, membrane-based NH 3 extraction from wastewaters represents a feasible technology for ammonia recovery over conventional methods (Brennan et al., 2020).An effective ammonia recovery can be achieved via membrane contactors as a result of their high surface area, selective NH 3 extraction and low consumption of energy (Mahmud et al., 2000;Yeon et al., 2003).NH 3 extraction from wastewater is based on NH 3 gas transfer between the wastewater and an acid solution separated by a hydrophobic membrane.The type and area of the membrane, which determines both the pores available and contact surface for NH 3 mass transfer, and the type of acid driving NH 3 diffusion, rank among the most important operational parameters of this technology (Guo et al., 2019;Hasanoĝlu et al., 2010;Rivera et al., 2022).Hollow fiber contactors are likely the most promising alternative for ammonia recovery in anaerobic broths.This configuration exhibits higher contact area per volume of membrane, allows a broad range of pHs, and exhibit a high resistance to chemicals and high temperatures, mechanical stability and hydrophobicity (Darestani et al., 2017;Eykens et al., 2016).Different chemical fertilizers such as ammonium sulfate, ammonium phosphate and ammonium nitrate can be produced depending on the type of acid used (Darestani et al., 2017).Despite this technology has been successfully tested to extract NH 3 from the effluent of anaerobic digesters, its direct implementation into an anaerobic digester to support in-situ ammonia recovery and enhance the anaerobic digestion process has never been tested.
This work investigated the potential of hydrophobic flat sheet membranes to in-situ and ex-situ extract NH 3 from the anaerobic broth of piggery wastewater digestion.Batch tests were initially conducted exsitu to study the influence of the concentration of suspended solids and pH on NH 3 extraction from anaerobically digested piggery wastewater using circular and rectangular configurations of Polytetrafluoroethylene PTFE hydrophobic membranes.The influence of in-situ NH 3 extraction from the culture broth of a continuous anaerobic digester treating piggery wastewater on the performance of anaerobic digestion was also investigated using the above cited membrane configurations.

Experimental Set-up
The first experimental set-up, composed of a continuous recirculation of anaerobically digested pig slurry (obtained from the second experimental set-up) using a peristaltic pump (Watson Marlow 520, USA) over the active layer of a PTFE hydrophobic membrane in two tailor made membrane holders, was used to investigate the influence of suspended solids and pH in the digested pig slurry on NH3 extraction.A similar peristaltic pump was used to recirculate a 1 M sulfuric acid solution on the support layer of the membrane.Both solutions were maintained in 0.3 L enclosed Erlenmeyer bottles kept in a thermostatic bath (HAAKE type E12, Thermo Fisher Scientific, Waltham, MA, USA) at 35 • C. The PTFE membrane was purchased from Millipore (Ireland) and exhibited a pore size of 0.22 µm, a nominal thickness of 170 µm, a contact angle of 150 • and a porosity of 70 %.This membrane was selected based on its effective performance during NH 3 recovery from sewage sludge digestates (Rivera et al., 2022).
The second experimental set-up was composed of a continuous stirred tank reactor (CSTR) with a working volume of 3L and magnetically stirred at 200 rpm in a thermostatic room (37 • C) (Fig. 1).The anaerobic cultivation broth present in the CSTR was tangentially recirculated at 0.25 L min − 1 using a peristaltic pump (Watson Marlow 520, Spirax-Sarco Engineering plc, Cheltenham, UK) over the active layer of a hydrophobic PTFE membrane casted on a circular (8.55 m 2 ) or rectangular (44. m 2 ) membrane holder (Silva et al., 2011).Ammonia was recovered in a 1 M sulfuric acid aqueous solution recirculated on the support layer of the membrane at 0.25 L min − 1 .Fresh piggery wastewater and anaerobic digestate were daily fed and withdrawn using peristaltic pumps (Watson-Marlow Sci-Q 323, Spirax-Sarco Engineering plc, Cheltenham, UK).

Influence of suspended solids and pH on NH 3 recovery from digested pig slurry
Four test series were carried out using the effluent from the 3L anaerobic digester treating piggery wastewater under mesophilic conditions.Two membrane holders with areas of 8.55 cm 2 (Montalvillo et al., 2014) and 44 cm 2 were used to investigate the influence of pH (8.2 ± 0.32 and 9.0 ± 0.05) and suspended solid concentration (raw digested pig slurry and digested pig slurry centrifuged at 10000 rpm for 10 min in a Thermo Scientific Sorvall TM Legend TM RT Plus centrifuge).The experiments were performed in duplicate at 35 • C using a 1 M solution of H 2 SO 4 .The recirculation rates of the digested pig slurries and H 2 SO 4 F. Rivera et al. solutions were set at 0.25 L min − 1 and the experiments lasted for 120 min.Samples (2 mL) for the determination of NH 3 concentrations were taken every 30 min.

Influence of membrane-based NH 3 extraction on the continuous anaerobic treatment of piggery wastewater
The experimental set-up was operated for 157 days under 3 different operational conditions.The bioreactor was inoculated with 2.5L of anaerobic inoculum, 0.5L of raw piggery wastewater and supplemented with 5 g NaHCO 3 L − 1 .Stage I was operated for 76 days with an average time spent by the substrate in the anaerobical digestor, hydraulic retention time (HRT), of 20 days without NH 3 extraction.Stage II involved the interconnection of the circular membrane holder at 0.25 L min − 1 with 1 M H 2 SO 4 for 17 days.Stage III involved the interconnection of the rectangular membrane holder at 0.25 L min − 1 with 1 M H 2 SO 4 for 64 days.The membrane was replaced weekly to maintain optimal NH 3 extraction performance.Samples (150 mL) of the raw piggery wastewater and the effluent of the anaerobic bioreactor were drawn twice a week to analyze the concentration of total ammoniacal nitrogen, total chemical oxygen demand (COD), total Kjeldahl nitrogen (TKN), total nitrogen (TN), total and volatile solids (TS, VS), total organic and inorganic carbon (TOC, IC) and volatile fatty acids (VFA), and temperature and pH.The volume and composition of the biogas daily produced were also recorded.

Analytical methods
Samples of biogas were taken using a gas-tight syringe (Hamilton, 1710 SL SYR, 100 µL, California, USA) for determining the concentration of CO 2 , CH 4 , H 2 S, O 2 and N 2 using a gas chromatograph (Varian CP-3800, Palo Alto, CA, USA) coupled with a detector and equipped with a CP-Molsieve 5 A (15 m × 0.53 mm × 15 µm) and a CP-Pora BOND Q (25 m × 0.53 mm × 15 µm) columns (García et al., 2019).The values of pH and temperature were monitored using a Basic 20 pH meter with a 50 14 T electrode (Crison Instruments, S.A., Barcelona, Spain) in the liquids of both sides of the membrane.Ammoniacal nitrogen was measured using the Nessler analytical method in a SPECTROstar Nano Absorbance Reader spectrophotometer at 425 nm (BMG LABTECH, Germany).The determination of COD, TKN, TS and vS was carried out according to Standard Methods for examination of water and wastewater (APHA, 2005).TN, TOC and IC were measured in a Shimadzu TOC-VCSH analyzer (Shimadzu, Kioto, Japan) equipped with a TNM-1 module.Volatile Fatty Acids (VFA) analysis was performed in an Agilent 7820A GC-FID (Agilent Technologies, Santa Clara, USA) equipped with a G4513A autosampler and a TEKNOKROMA NF29370-F packed column (2 m × 1/8 ′′ × 2.1 mm) (Teknokroma, Barcelona, Spain) (López et al., 2018).

Influence of suspended solids and pH on NH 3 recovery from digested pig slurry
The increase in pH from 8.2 to 9 in the digestate resulted in an increase in the NH 3 extraction efficiency regardless of the area of the membrane and presence of suspended solids.Thus, an increase in the NH 3 concentration removed in 2 h from 0.20 and 0.21 g/L at pH 8.2 up to 0.28 and 0.31 g/L at pH 9 was recorded in the presence and absence of suspended solids, respectively, in the 8.55 cm 2 circular membrane module.In the 44 cm 2 rectangular membrane module, the NH 3 concentration removed in 2 h increased from 0.31 and 0.33 g/L at pH 8.2 up to 0.40 and 0.43 g/L at pH 9 in the experiments carried out with raw and centrifuged digestate, respectively.Thus, the removal efficiency of ammonia with a previous centrifugation at pH 8.2 accounted for 15.85 %, while at pH 9 this removal efficiency increased up to 20.94 %.The pH of the digestate is a key parameter during NH 3 permeation in membranes since higher pHs swift the NH 4 + /NH 3 equilibrium in the digestate towards NH 3 , the chemical compound that is able to permeate through the membrane, which results in higher ammonia removal rates (Table 1).Previous works identified an optimal pH value of 9 for NH recovery from real digestates along with other operational parameters of the PTFE membrane module such as a 1 M sulfuric acid concentration and recirculation flow rates of 0.25 L min − 1 (Molinuevo- Salces et al., 2020;Rivera et al., 2022).Unfortunately, pH values higher than 9 in the anaerobic broth of digesters could negatively impact the structure and performance of the prevailing microbial anaerobic community, which would ultimately deteriorate waste treatment performance and biogas production (see supplementary material).
The circular and rectangular membrane modules were compared in terms of the influence of the flow hydrodynamics on NH 3 molar fluxes.Thus, molar fluxes of 3.15 mol m -2 h − 1 and 0.85 mol m -2 h − 1 were recorded using solid-free digestate at pH 9 in the 8.55 cm 2 and 44 cm membrane modules, respectively.These values were in agreement with  the molar flux of 4.52 mol m -2 h − 1 reported for synthetic anaerobic digestate at pH 10 under similar operating conditions with a circular PTFE membrane module (Rivera et al., 2022).The speed of the fluid within the circular module is 1.61 m s − 1 with a Reynolds number of 1,640, while for the rectangular module the values were 0.10 m s − 1 and 106 respectively.For both modules the fluid flow regime was laminar due to the low-velocity fluids.However, through the circular module the Reynolds numbers is 15 -fold higher than the rectangular and this explains the difference in the NH 3 molar flux values.
In brief, this study revealed that greater concentrations of NH 3 were recovered when increasing the membrane surface and pH, while the presence of suspended solids did not hinder ammonia extraction significantly in these short-term experiments.This latter finding represents a competitive advantage from an operational point of view (no digestate centrifugation needed), although some suspended solids would end up deposited on the membrane surface (fouling) during long-term experiments and limit NH 3 recovery.

Influence of membrane-based NH 3 extraction on the continuous anaerobic treatment of piggery wastewater
Since NH 3 is one of the mayor inhibitors of the anaerobic treatment of livestock wastewaters (Müller et al., 2006), the reduction of Total Ammonia Nitrogen (TAN) concentrations below inhibitory levels but above nutritional requirements is expected to enhance the performance of anaerobic digestion in terms of COD and vS removal, and ultimately of biogas productivities.In this context, (Siles et al., 2010) reported a NH 3 threshold of 0.62 g free ammonia L − 1 during the anaerobic digestion of a synthetic solution rich in ammonia and sodium sulphate, with a decrease of 21 % in biogas production observed when the concentration of NH 3 was increased.Other authors have reported inhibitory TAN concentration ranges of 1.5 -7.0 g N L − 1 in anaerobic digesters (Hejnfelt and Angelidaki, 2009).The TAN inhibition threshold in anaerobic cultivation broths depends on the type and concentration of substrates, pH, temperature, microbial population structure, etc. (Chen et al., 2008).It is typically assumed that TAN concentrations above 3.0 g NH 4 -N L − 1 causes a severe inhibition of the methanogenesis of the process   ( McCarty, 1964).
Temperature and pH are one of the most important parameters governing ammonia removal since they determine the rate of ammonia mass transfer across the membrane.Indeed, a temperature increase of 30 ªC entailed an increase in ammonia mass transfer by a factor of 7.3-fold using synthetic wastewaters (Deĝermenci et al., 2012;Jiang et al., 2014).
The anaerobic treatment of centrifuged piggery wastewater resulted in steady state TKN and TN concentrations of 4.77 ± 0.04 and 4.01 ± 0.25 gN L − 1 during stage I (Fig. 2a).The implementation of NH 3 extraction using the circular flat sheet membrane induced a decrease in TKN and TN concentrations to 4.45 ± 0.07 g N L − 1 and 3.64 ± 0.17 gN L − 1 by the end of stage II, and to 3.77 ± 0.01 g N L − 1 and 3.38 ± 0.02 gN L − 1 by the end of stage III, respectively (Fig. 2a; Fig. 3c).There was also a noticeable decrease in NH 3 concentrations from 1.58 ± 0.05g NH 3 L − 1 at the end of stage I, to 1.50 ± 0.06 g NH 3 L − 1 at the end of stage II and to 1.18 ± 0.05 g NH 3 L − 1 at the end of stage III (Fig. 2b).The rates of ammonium removal in stages I, II and III accounted for 3.99 mg L − 1 d − 1 , 8.81 mg L − 1 d − 1 and 24.61 mg L − 1 d − 1 , respectively.In this context, the molar fluxes of NH 3 across the PTFE membrane in stages II and III averaged 0.41 mol NH 3 m 2 h − 1 at pH 8.15 ± 0.09 and 0.05 mol NH m 2 h − 1 at pH 8.22 ± 0.02, respectively.These molar fluxes were much lower than those recorded in the batch tests at pH values of 8.2 ± 0.32 with raw digestate (2.06 mol NH 3 m -2 h − 1 and 0.61 mol NH 3 m -2 h − 1 for the circular and rectangular membrane modules).The lower performance of the membrane modules under long term operation could be explained by the gradual membrane fouling, which hindered NH permeation through the membrane.
Membrane fouling is caused by microorganisms, organic and inorganic elements.According to extensive studies it was detected that fouling layers cause membranes to lose hydrophobicity, which also leads to reducing the period of efficiency (Xu et al., 2010;Zarebska et al., Fig. 3. Time course of the TOC, IC and TN concentrations of the feed (•) and effluent (▴) during the three operational stages.
F. Rivera et al. 2014).To reduce this issue, some physical and chemical cleaning methods could be applied (Darestani et al., 2017).According to (Chang et al., 2002), it is possible to achieve a complete abatement for fouling and blocking problems applying both methods and making a periodic application of chemicals.
To the best of the authors knowledge, this work represents the first study reporting in-situ NH 3 extraction using an external membrane module interconnected to a continuous anaerobic digester.This process configuration requires less maintenance, low operational costs and can support a high ammonia recovery.When the pH in the anaerobic cultivation broth was increased, NH 3 removal increased since the equilibrium is shifted towards NH3.The un-ionized NH 3 permeates through the pores to the support layer, where the acid solution is.When NH 3 reaches the acidic solution, it forms NH 4 + ions by combining with free protons, thus maintaining the maximum NH 3 concentration gradient across the membrane (Dube et al., 2016;Garcia-González and Vanotti, 2015).COD and vS removal efficiencies during Stage I averaged 33.05 % ± 2.01 % and 25.67 % ± 2.32 %, respectively, under steady state (Fig. 4).The removal efficiencies of COD during Stage II accounted for 42.88 % ± 2.57 %, while a significant increase up to steady state removals of 61.80 % ± 1.33 % was observed during Stage III, likely due to the reduction in NH 3 toxicity caused by the in-situ NH 3 extraction implemented (Fig. 4a).Similarly, vS removals during Stage II and III increased up to 26.83 % ± 3.01 % and 37.89 % ± 1.84 % as a result of the reduction in NH 3 inhibition (Fig. 4b).Other studies treating swine manure diluted with water (at an initial COD of 34.16 g/L) under mesophilic conditions in a semi-continuous CSTR operated at a HRT of 5 days reported COD removal efficiencies of 58.26 %.These authors reached COD removal efficiencies of 68.8 % when digesting swine manure pre-treated with a tubular PTFE gas membrane (Molinuevo-Salces et al., 2018).Other authors have observed that a reduction in TKN can increase COD removal by 55 %, which directly enhances the assimilation of VFA.However, methanogenesis does not take place when NH 3 concentration decreases from 0.1 g to 0.01 g NH 4 -N L − (Panichnumsin et al., 2010;Resch et al., 2011).
CH 4 yields of 380.4 ± 85.0, 542.2 ± 52.8 and 566.1 ± 47.8 NmLCH g vS fed − 1 were recorded during Stage I, II and III, respectively (Fig. 5a).These methane yields matched the observed trends in COD and vS removals along the three operational stages.In this context, (Hansen et al., 1999) reported values of 188 mLCH 4 g vS fed − 1 and 65 % CH 4 content in a mesophilic CSTR bioreactor fed with fresh swine manure at 20 days of HRT.(González-García et al., 2021) compared two anaerobic reactors, with and without an internal gas permeable membrane, operating under similar conditions using swine manure as a feedstock, and reported an improvement in the anaerobic digestion process with the recovery of NH 3 .Thus, an increase in the CH 4 content of biogas of up to 14 % and in the methane yield of up to 9 % higher (70 NmLCH 4 g vS fed − 1 ) was recorded in the digester equipped with an internal membrane module.
Biogas compositions remained stable during stages I, II and III, with concentrations of CO 2 of 20.75 % ± 1.22 %, 20.67 % ± 0.01 % and 23.24 % ± 0.29 %, respectively (Fig. 5b).CH 4 contents of 77.65 % ± 1.08 %, 77.16 % ± 2.42 % and 75.11 % ± 0.01 % were also recorded in stages I, II and III (Fig. 5c), respectively.At this point it should be highlighted that the values of the anaerobic broth pH during stages II and III varied as a result of the weekly membrane replacement, which entailed fluctuations in the CO 2 and CH 4 contents in the biogas.Indeed, the use of new membranes resulted in an improved NH 3 removal from the anaerobic broth, but also proton transfer from the 1 M H 2 SO 4 solution to the anaerobic broth.
Finally, the VFA concentrations present in the anaerobic broth decreased significantly as a result of membrane operation.Thus, the removal efficiencies of acetic acid accounted for 72.0 % ± 8.1 %, 73.4 % ± 0.2 % and 93.7 % ± 1.9 %, respectively, during stages I, II and III.Similarly, the removals of propionic acid remained constant during stages I and II at 0 % ± 0 %, 0.5 % ± 0.8 %, respectively, but increased up to 98.6 % ± 1.9 % in stage III.A stepwise increase was also observed in isobutyric acid removals in stages I, II and II (48.3 % ± 4.8 %, 80.1 % ± 3.1 % and 100 % ± 0 %, respectively).The removals of butyric acid increased from 80.6 % ± 24.6 % in stage I up to 97.2 % ± 0.9 % and 100 % ± 0 % in stages II and III, respectively.Finally, the removals of isovaleric acid accounted for 61 % ± 31.3 %, 53.2 % ± 0.4 % and 100 % ± 0 % and those of valeric acid for 50.0 % ± 30.7 %, 78.4 % ± 3.5 % and 100 % ± 0 % in stages I, II and III (Fig. 6).These enhanced VFA removal efficiencies were likely explained by the enhancement in the anaerobic digestion process mediated by the gradual extraction of NH 3 from the anaerobic broth.At this is point it should be highlighted that a potential VFA mass transfer from the anaerobic broth to the 1 M H 2 SO 4 solution was ruled out because concentrations of VFAs were low, which makes difficult to achieve a liquid contact through the pores of the membrane.The implementation of membrane-based NH 3 extraction gradually reduced the propionate/acetate ratio in the anaerobic broth, which accounted for 1.39, 1.29 and 0.81, respectively.In this context, propionate/acetate ratios > 1.4 are known to deteriorate anaerobic digestion performance (Gao et al., 2015), since propionic acid is considered the most toxic volatile fatty acid (Wang et al., 2009).Overall, high VFA F. Rivera et al. concentrations can cause inhibition to the methanogens during swine manure digestion along with a decrease in the buffer capacity and pH of the cultivation broth (Axelsson et al., 2012;Kraemer and Bagley, 2007).(Hansen et al., 1998) concluded that high VFA concentrations caused inhibition of the anaerobic process with a low methane yield of 67 mLCH 4 g vS − 1 during the operation of a CSTR treating swine manure at 25 days HRT.In our particular study, these low VFA concentrations induced by NH 3 extraction further enhanced the anaerobic digestion of swine manure.

Conclusions
The anaerobic treatment of piggery wastewater can be enhanced by reducing NH 3 concentrations in the anaerobic broth via membrane assisted NH 3 extraction.Ammonia recovery increases at increasing pH values in the digestate.The concentration of suspended solid in the anaerobic broth did not significantly influence NH 3 extraction under short term operation.NH 3 extraction from the anaerobic digester during continuous pig slurry treatment supported enhancements in COD and vS removal efficiencies of 87 % and 48 %, respectively, and mediated a complete VFA assimilation.In addition, while biogas composition remained constant, the yields of CH 4 increased by 49 %.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.Schematic representation of the CSTR anaerobic digester coupled with an external membrane-based NH 3 extraction unit.

F
.Rivera et al.

Fig. 2 .
Fig. 2. Time course of the TKN (a) and NH 3 (b) concentrations in the feed (•) and anaerobic effluent (▴) during the three operational stages.

F.
Rivera et al.

Fig. 4 .
Fig. 4. Time course of the concentrations of COD (a) and vS (b) in the feed (•) and effluent (▴), and their corresponding removals ( ) during the three stages of operation.

Table 1
Operational conditions and NH 3 recovery indicators in batch extraction assays.