Electrochemical Ammonia Recovery from Source-Separated Urine for

Nov 7, 2017 - Energy savings through accompanying electrolytic H2 and O2 production were accounted for. Subsequently, MP was grown in fed-batch on MP ...
1 downloads 4 Views 1MB Size
Download PDF
Article pubs.acs.org/est

Cite This: Environ. Sci. Technol. XXXX, XXX, XXX-XXX

Electrochemical Ammonia Recovery from Source-Separated Urine for Microbial Protein Production Marlies E.R. Christiaens, Sylvia Gildemyn,† Silvio Matassa,‡ Tess Ysebaert, Jo De Vrieze, and Korneel Rabaey* Center for Microbial Ecology and Technology (CMET), Ghent University, Coupure Links 653, B-9000 Gent, Belgium S Supporting Information *

ABSTRACT: Conventional plant and meat protein production have low nitrogen usage efficiencies and high energy needs. Microbial protein (MP) is an alternative that offers higher nitrogen conversion efficiencies with low energy needs if nitrogen is recovered from a concentrated waste source such as source-separated urine. An electrochemical cell (EC) was optimized for ammonia recovery as NH3/H2 gas mixtures usable for MP production. Undiluted hydrolyzed urine was fed to the caustic-generating cathode compartment for ammonia stripping with redirection to the anode compartment for additional ammonium extraction. Using synthetic urine at 48 A m−2 the nitrogen removal efficiency reached 91.6 ± 2.1%. Tests with real urine at 20 A m−2, achieved 87.1 ± 6.0% and 68.4 ± 14.6% requiring 5.8 and 13.9 kWh kg N−1 recovered, via absorption in acid or MP medium, respectively. Energy savings through accompanying electrolytic H2 and O2 production were accounted for. Subsequently, MP was grown in fed-batch on MP medium with conventional NH4+ or urine-derived NH3 yielding 3.74 ± 1.79 and 4.44 ± 1.59 g CDW L−1, respectively. Dissolution of gaseous NH3 in MP medium maintained neutral pH in the MP reactor preventing caustic addition and thus salt accumulation. Urine-nitrogen could thus be valorized as MP via electrochemical ammonia recovery.

■

INTRODUCTION Newly built wastewater treatment plants are still based on the 100-year old activated sludge system with focus on removal rather than recovery. Nitrogen removal in these systems comes at high energy costs of 4.4 to 11.1 kWh kg N−1 which could be avoided if the nitrogen was recovered.1 Simultaneously, ammonia production via the conventional Haber-Bosch process accounts for 1−2% of worldwide energy consumption (10.3− 12.5 kWh kg N−1)1 providing 70% (100 Mt) of nitrogen fertilizer needs.2 Usage efficiencies of fertilizer nitrogen in plant3 and meat protein production are as low as 9% (13 Mt) and 7% (10 Mt) of total fertilizer, respectively,2 due to losses from runoff,4 denitrification,3 or NH3 volatilization.5 Microbial protein (MP) is a more efficient and high-rate alternative, reemerging as a fertilizer, feed, or food additive which can be produced heterotrophically or autotrophically.6,7 For example, hydrogen oxidizing bacteria (HOB) are chemolithoautotrophic bacteria containing up to 71% crude protein (12% N) based on cell dry weight (CDW).8,9 HOB can be grown on waste CO2, renewable energy sourced electrolytically produced H2 and O2, and typically NH4+ as nitrogen source.9,10 This could yield potentially 3120 ton protein ha−1 yr−1 compared to conventional soy yields of 3 ton ha−1 yr−1.2 Consequently, MP could be part of the solution to feed an ever-growing world population while releasing pressure on arable land. Wastewater treatment and food production could thus come together via ammonia, but the typically low total ammonia © XXXX American Chemical Society

nitrogen (TAN) concentration in wastewater makes harvesting this nitrogen a key challenge. A solution is source-separation in conjunction with waterless collection of urine. Urine constitutes only 1% of domestic wastewater by volume in developed countries but comes with 4−14 g N L−1 of which up to 90% is urea.11 Ureolytic bacteria are ubiquitous,12 quickly catalyzing TAN production from urea.13 Especially at airports, shopping malls, or in the megacities of the future, urine-nitrogen availability warrants direct recovery. Several technologies have been studied to recover TAN directly from urine. Thus, far, none seemed to be competitive with Haber-Bosch due to e.g. economics of scale. Direct gas stripping of hydrolyzed urine is an established technique to recover TAN as (NH4)2SO4 in an absorption column,15,16 with TAN removal efficiencies up to 98.7%.14 However, alkali addition to pH 9.5−12 and/or heating to 16−85 °C results in a high chemical and energy demand (3.9 to 28.2 kWh kg N−1 operation- and scale-dependent).14,15,17−19 More recently, (bio)electrochemical systems combined with stripping and absorption have been developed. These systems consist of an anode generating protons and a cathode generating hydroxyl anions separated by an ion exchange membrane. Hydrolysed Received: June 1, 2017 Revised: October 12, 2017 Accepted: October 23, 2017

A

DOI: 10.1021/acs.est.7b02819 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

Figure 1. Reactor setup for continuous nitrogen stripping and extraction from urine combined with H2, O2, and CO2 production in an electrochemical cell. Full lines represent liquid streams, dashed lines gas streams, and x indicates sampling points.

■

MATERIAL AND METHODS Setup for Nitrogen Recovery and Gas Production. Nitrogen stripping and extraction were achieved by coupling the cathode compartment of a two-chamber electrochemical cell to a stripping and an absorption column, while redirecting the stripped catholyte to the anode compartment (Figure 1). The cell was built as previously described,21 with an inner compartment size of 5 × 20 × 2 cm3 and 100 cm2 projected surface area for electrodes and the cation exchange membrane (CEM). A Ag/AgCl reference electrode (±0.247 V vs standard hydrogen electrode, 3 M NaCl, Bio-Logic, France) was placed in the cathode compartment. A VSP 5-channel galvanostat (Bio-Logic, France) controlled current and measured cathode potential and cell voltage. Synthetic or real hydrolyzed urine was continuously fed in the cathode compartment and recirculated over a tubular stripping column (6 × 108 cm2). The counter gas flow was recirculated over a tubular absorption column allowing ammonia to dissolve in a known volume of 3 M H2SO4. Excess gas was quantified by means of a calibrated gas counter (Bakker&Co, Belgium). A water trap was installed to protect the gas pump. From the catholyte recirculation loop liquid was fed in the anode compartment and recirculated. Anolyte effluent was passively collected in a gastight 2L glass bottle. Excess gas moved to the headspace and was quantified with a second gas counter. A detailed material description can be found in Supporting Information (SI) S1. Urine. The reactor concept was characterized with synthetic urine according to a recipe modified after Bonvin et al.25 (SI S2 Table), based on a preliminary characterization of hydrolyzed morning real male urine. Validation of the reactor concept was done with real urine. Batches of fresh male urine were collected throughout the day at Ghent University with permission of the university ethical committee and autohydrolysed in a closed vessel at 28 °C until use, although ureolysis occurred within days. Lower salinity (42.6 ± 7.4 vs 106.1 ± 8.9 mS cm−1) and TAN concentrations (5.49 ± 0.53 vs 8.60 ± 1.01 g N L−1) were measured for hydrolyzed real compared to synthetic urine (SI S3 Table) indicating the time-dependent variability in real urine. Before connecting real hydrolyzed urine to the cathode compartment, a N2 filled gas bag (KeikaVentures, NC) was attached to the feed vessel headspace to keep it anaerobic upon

urine enters the anode compartment while current, either produced from microbial organics oxidation (microbial fuel cell, MFC) or (partly) applied (microbial electrochemical cell, MEC or electrochemical cell, EC), pulls positively charged molecules, mainly NH4+, over the membrane.20,21,23,24 In the alkaline cathode compartment gas stripping removes NH3 without the need for caustic dosing or heating.21−24 At the anode, CO2 is produced from organics oxidation in MFC/MEC while O2 is produced in EC. Both MEC and EC configurations require energy input, enabling generation of H2 gas at the cathode, which can be used as a fuel to recover part of the electrical energy applied. In an EC,21 this led to energy investments of 8 kWh kg N−1 when the energy value of the coproduced H2 was considered, which is lower than Haber-Bosch. However, with a maximum TAN removal efficiency of 75% this technology cannot yet reach similar nitrogen removal efficiencies or the lowest energy requirement levels of conventional stripping and absorption. We hypothesize higher nitrogen removal efficiencies could be achieved with an EC by introducing hydrolyzed urine first in the cathode compartment for alkalifying and stripping, then redirecting the effluent to the acidic anode compartment for additional electrochemical NH4+ extraction. H2 and NH3 from the cathode and O2 and CO2 (originating from the urea hydrolysis product HCO3−) from the anode can then be used for microbial protein production by HOB. In this way, nitrogen is recycled directly preventing costs for conventional nitrogen removal and production by Haber-Bosch, and CO2 is sequestered in MP. Use of gases as substrate can eliminate potential pathogen and micropollutant crossover from urine to MP. This study presents the proof-of-concept for this approach, first with synthetic urine to quantify removal efficiencies and gas production rates subject to operational parameter changes, and second with real, hydrolyzed male urine. Ammonia recovered in HOB medium is fed to a HOB production reactor and compared with NH4Cl as a conventional nitrogen source. B

DOI: 10.1021/acs.est.7b02819 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

Table 1. Overview of Electrochemical Experiments with Variable Operational Parameters (Average ± SD, n = 3). Tests Were Run in Duplicate Except * test ID: urine and absorbent a

Syn12 acid Syn34 acida Real acid Real medium

j (A m−2) b

0.1 , 20, 30, 48 0.1b, 20, 30, 48 0.1b, 10*, 20 20

HRT cathode (h) 9.4 10 8.3 8.3

± ± ± ±

0.2 0.2 0.0 0.0

HRT anode (h) 5.8 7.4 5.7 5.7

± ± ± ±

0.4 0.5 0.1 0.1

Qgas (L min−1) 1.4 7.1 7.1 7.1

± ± ± ±

0.1 0.0 0.0 0.0

Qrecirc (L d−1) 119 120 306 306

± ± ± ±

1 5 84 84

a

Syn12 acid refers to duplicate experiments 1 and 2 with synthetic urine and 3 M H2SO4 as absorbent, syn34 acid refers to duplicate experiments 3 and 4. bControl: A current density of 0.1 A m−2 did not allow electromigration but only diffusion over the CEM. Hence, this operational condition was considered as only stripping.

a 930 Compact IC Flex with a Metrosep A supp 5 guard and A supp 5 150/4.0 main column, and a 930 Compact IC Flex with a Metrosep 4.6 guard and 250/7.8 main column, respectively, all equipped with conductivity detectors (Metrohm, Switzerland). HOB biomass was quantified as total suspended solids (TSS) after centrifugation at 12 000 rpm for 10 min and washing.26 Liquid biomass samples for community analysis were centrifuged at 12 000 rpm for 10 min at the start and end of HOB fed-batch tests. The pellets were stored at −20 °C for total DNA extraction.27 PCR was conducted (5′ at 94 °C, 30 cycles of consequently 1′ at 95 °C, 1′ at 53 °C, 2′ at 72 °C, and a final 10′ at 72 °C) with 63F and 1387R primers (Biolegio, Nijmegen, The Netherlands).28 The quality of the DNA extracts (clear, unfragmented band) and PCR products (single band with the targeted length) was validated by 1% agarose gel electrophoresis. PCR products were purified with the innuPREP PCR pure kit (Analytikjena, Jena, Germany) before sanger sequencing (LGC Genomics GmbH, Germany). Results were blasted via NCBI and submitted in the European Nucleotide Archive (ENA; accession number PRJEB22958). Calculations. Relevant calculations performed are described in SI S6. Units expressed as N refer to TAN, unless stated otherwise.

volume reduction. To avoid foaming of real urine in the system, 1 mL L−1 of antifoaming agent (Schill+Seilacher “Struktol” GmbH, Germany) was added to the feed vessel. Experimental Design and Rector Operation. Apart from the difference in feed (synthetic and real hydrolyzed urine), the following operational parameters were varied: current density (j, A m−2), hydraulic retention time (HRT, d), liquid (Qrecirc, L d−1) and gas (Qgas, L min−1) recirculation rates, and the liquid in the absorption column, that is, 3 M H2SO4 or HOB medium (Table 1). All experiments were run in duplicate except for real acid 0.1 and 10 A m−2. Samples were taken during steady state, that is, four HRTs after a parameter was changed. The feed, catholyte, anolyte, stripping column effluent, and absorbent (Figure 1) were sampled and analyzed for gas composition, pH, electrical conductivity, cations, and anions. Averages and standard deviations (SD) were calculated from at least three steady state data points over duplicate tests. Experiments were conducted at 23 ± 2 °C. Microbial Protein Production. The production of microbial protein with urine-derived NH3 dissolved in HOB medium was demonstrated and compared with conventional NH4Cl as a control. A bubble column (6 × 108 cm2, 0.7L working volume, Exclusief Glas, The Netherlands) was inoculated at 1 g TSS L−1 with an enriched HOB culture, 97% dominated by Sulf uricurvum spp.,9 and operated in fed-batch mode for 10 days (detailed operational parameters: SI S4 Table). Nitrogen was not added via the HOB medium (SI S5 Table) but separately spiked as NH4Cl or urine-NH3 recovered in HOB medium, if the concentration decreased below 0.25 g N L−1. H2 and O2 produced via two electrolyzers (detailed description in SI S6) and CO2 from a gas bottle (Lindegas, Belgium) were continuously dispersed and recirculated in the column via a sintered glass frit (160−250 μm pores) at the bottom. Excess gas was quantified by a calibrated gas counter (Bakker&Co, Belgium). Biomass growth, TAN depletion, electrical conductivity (Metrohm, Belgium), and pH (Metrohm, Belgium) were followed on daily basis. Only for the test with NH4Cl, pH was manually adjusted at 6.5 ± 0.5 by dosing 2 M NaOH. Chemical and Molecular Analyses. Gas composition was quantified by compact gas chromatography (Global Analyzer Solutions, Breda, The Netherlands) equipped with a Molsieve 5A precolumn and Porabond column (O2, N2, CH4, H2) and a Rt-Q-bond precolumn and column (CO2, N2O). Total and soluble (0.22 μm filtered) chemical oxygen demand (tCOD and sCOD), carbonaceous biological oxygen demand over 5 days (cBOD5), total Kjeldahl nitrogen (TKN), pH, and electrical conductivity were analyzed according to standard methods.26 Filtered samples (0.22 μm) were analyzed for cations (Na+, TAN, K+), anions (Cl−, NO2−, NO3−, PO43−, SO42−), and volatile fatty acids (VFA) by a 761 Compact IC with a Metrosep C4/4.0 guard and C6−150/4.0 main column,

■

RESULTS Nitrogen Removal from Synthetic Urine. Synthetic hydrolyzed urine was fed in the cathode of an EC coupled to a stripping and an absorption column containing H2SO4 to characterize TAN removal. Applying increasing current densities increased TAN removal from 41.1 ± 8.4% and 45 ± 12.1% at stripping conditions (0.1 A m−2) for synthetic urine with low (syn12) and high (syn34) gas recirculation rates, respectively, to 74.4 ± 4.2% and 91.6 ± 2.1% at 48 A m−2 (Figure 2; SI S7 Table). This increase in nitrogen removal could be attributed to increasing OH − and thus NH 3 production at the cathode (SI S9 Figure). The lower nitrogen removal for syn12 compared to syn34 might be attributed to a decreased NH3 absorption capacity of H2SO4 during tests with low gas recirculation rate (syn12) (SI S8 Figure). NH3 Stripping Drives NH4+ Migration Over the CEM toward the Cathode. The rationale for passing the catholyte effluent to the anode compartment was additional removal of NH4+ over the CEM besides neutralizing the effluent pH. At 48 A m−2, 55.8 ± 7.8% and 68.9 ± 9.1% of the remaining catholyte TAN was removed for syn12 and syn34, respectively (SI S7 Table). As a result, effluent TAN concentrations decreased to 2.5 ± 0.2 and 1.3 ± 0.3 g N L−1. Cathodic TAN showed an increasing NH3 fraction with applied current density because electrochemical OH− production increased the catholyte pH to or above the pKa of NH4+/NH3, which is 9.3 ± 0.05 at 23 ± 2 °C (SI S9 Figure). Remaining catholyte TAN was converted to C

DOI: 10.1021/acs.est.7b02819 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

0.003 M) in the bulk liquid. As a CEM was used, a potential anion flux is negligible compared to the cation flux, as was proven by equal anion concentrations in anolyte and catholyte (data not shown). NH4+ mainly diffused to the cathodic compartment as the continuous removal of NH3 via stripping pulled the NH4+/NH3 equilibrium, established by the applied current density and thus catholyte pH, toward NH3. Catholyte and anolyte NH4+ concentrations were indeed in equilibrium irrespective of the current density. Moreover, at stripping conditions (0.1 A m−2) 26.2 ± 6% and 22.1 ± 12.2% of catholyte TAN was removed in the anodic compartment for syn12 and syn34, respectively (S7 Table). Catholyte and anolyte pH were similar as the applied current density was too low to generate a pH gradient. Consequently, the same ratio NH4+/NH3 was maintained in both compartments resulting in roughly 25% nitrogen extraction due to NH3 stripping. Nitrogen Removal and Extraction Mechanism with Real Urine. Real hydrolyzed urine was fed in the cathode compartment to validate the results achieved with synthetic urine. Nitrogen removal reached 87.1 ± 6.0% at 20 A m−2 for real urine with acid absorption, which was similar to synthetic urine at 48 A m−2 (Figure 2). The lower TAN concentration for real urine (SI S3 Table) and thus buffering capacity combined with enhanced mixing due to a doubled liquid recirculation rate (Table 1) resulted in cathodic pH values and an effluent TAN concentration (1.15 ± 0.69 g N L−1) similar to synthetic urine at 48 A m−2 (SI S7 Table). Diffusion of NH4+ over the CEM toward the cathode compartment removed 42.7 ± 29.2% nitrogen from the anolyte (Figure 3). Nitrogen removal in real urine with NH3 absorption in HOB medium at 20 A m−2 was similar to synthetic urine at 20 A m−2. The positive effect of the increased cathodic pH was countered by the lower absorption capacity of the HOB medium compared to H2SO4, leading to NH3 accumulation in the cathode compartment gas and liquid phase (SI S9 Figure). Still, NH4+ diffusion to the cathodic compartment occurred. Although the alkaline anolyte pH established equal concentrations of NH3 and NH4+, these were higher than NH4+ in the cathode compartment resulting in 60.7 ± 22.1% nitrogen diffusion over the CEM. Interestingly, compared to real urine and acid absorbent, less nitrogen removal by stripping and absorption resulted in more relative nitrogen diffusion over the CEM, probably due to the higher catholyte pH and compartments concentration gradient. Despite the alkaline anodic pH no NH3 was lost to the gas phase shown by equal anodic and feed TAN concentrations (SI S9 Figure). Furthermore, the theoretical charge balance was closed by Na+ and K+ basically excluding electromigration of NH4+ (Figure 3). NH3 Recovery and Gas Production for Microbial Protein Production. Operating the electrochemical cell at 20 A m−2 with real urine and HOB medium as absorbent resulted in an NH3 recovery of 0.05 ± 0.01 mol N d−1 or 0.48 ± 0.05 mol N L−1 (SI S7 Table). An overall nitrogen mass balance is provided in SI S11 Figure. Recovery efficiency reached 13.3 ± 3.3%. This recovered nitrogen could yield 6.2 g HOB-CDW d−1 in fed-batch requiring 7.5 mol H2 d−1, 2.4 mol O2 d−1, and 1.6 mol CO2 d−1.9 At 20 A m−2, however, these gas quantities could not be produced. At the cathode, 0.03 ± 0.01 mol H2 d−1 originated from water reduction corresponding to 28.5 ± 16.1% CE. At the anode, water oxidation yielded 0.03 ± 0.01 mol O2 d−1 compared to the theoretical 0.045 mol d−1 (SI

Figure 2. Total ammonium nitrogen (TAN) removal (%) increases with catholyte pH induced by current density (j, A m−2). Averages (±SD) were calculated for at least 3 steady state data points over duplicate tests, except for real acid 0.1 and 10 A m−2.

NH4+ in the acidifying anodic compartment (pH 5.1 ± 1 and 1.3 ± 0.1 for syn12 and 34, respectively). This acidification allowed electromigration or at least diffusion of NH4+ over the CEM toward the cathodic compartment. Nitrogen loss toward the effluent gas phase or as precipitation was implausible given the acid anolyte. Coulombic efficiencies (CE) suggest that mainly Na+ and K+ made up 100% of the charge balance (Figure 3). This was confirmed by their movement against

Figure 3. Stacked Coulombic efficiencies as a function of current density (j, A m−2) show TAN diffusion over the CEM while Na+ and K+ cross by electromigration as they fit the theoretical charge balance (100% CE). Averages (±SD) were calculated on at least three steady state data points over duplicate tests, except for real acid 10 A m−2.

concentration gradients, proving electromigration (depleting anolyte and increasing catholyte concentrations in SI S9 Figure). Protons were not expected to take a considerable part in charge balancing because anolyte concentrations were too low compared to Na+ and K+ (SI S9 Figure). Only for syn34 at 48 A m−2, protons reached similar K+ concentrations (0.05 ± D

DOI: 10.1021/acs.est.7b02819 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology S10 Figure), achieving a CE of 61.2 ± 32.7%. Microbial biomass attached to the anode probably consumed O2 resulting in oxidation of 57 ± 45% of feed VFA-COD. In addition, HCO3− conversion to CO2 was limited because the anolyte pH (9.1 ± 0.6) was well above 6.37, the pKa of H2CO3/HCO3−. Operation at 20 A m−2 did not show anodic Cl2 production although the anode potential (1.36 ± 0.03 V vs Ag/AgCl) was above 1.21 V vs Ag/AgCl (i.e., the standard reduction potential for chloride oxidation at pH 9, 25 °C, 1 bar, and measured chloride concentrations (SI S3 Table)). Production of OCl− could not be excluded, but it could not be converted to Cl2 gas due to the alkaline anolyte pH. Moreover, observed anodic biomass growth indicated that at least no toxic levels of chlorine species were produced. For MP production, an additional 20 A electrolyzer (assuming 83% CE) and CO2 source would be needed to achieve the optimal gas ratios. Electrochemically Recovered NH3 Did Not Limit Growth of HOB. Microbial protein production on urine as a nitrogen source was demonstrated by growing HOB on urinerecovered TAN and NH4Cl as a control at the same loading rate (SI S4 Table). For safety and lab-scale reasons, HOB were fed 16% of daily recovered TAN (0.008 mol N d−1) to limit the electrolyzer to 900 mA for H2 and O2 production. Both tests showed the same dominance by Sulf uricurvum spp. as the inoculum proven by sanger sequencing. Fed-batch growth stabilized by day 6 at 5.4 ± 0.6 and 5.6 ± 0.3 g CDW L−1 for urine-derived TAN and NH4Cl, respectively (SI Figure S12) and maximal production rates were 1.25 and 1.13 g CDW d−1, respectively. Biomass concentrations and TAN removal efficiencies (Table 2) were two times lower compared to

electrochemical cells, TAN removal from hydrolyzed urine improved by first passing through the cathode compartment with immediate NH3 stripping before redirecting the catholyte to the anode for additional effluent polishing via NH4+ diffusion over the CEM. Removal efficiencies up to 91.6 ± 2.1% (0.94 ± 0.03 mol N L−1 d−1) for synthetic urine at 48 A m−2 and 87.1 ± 6% (0.57 ± 0.07 mol N L−1 d−1) for real urine at 20 A m−2 have, to our knowledge, not been reported before in (M)ECs that fed hydrolyzed urine to the anode and relied on NH4+ electromigration over the CEM. Recently, Arrendondo et al. (2017)36 reported 92% removal efficiency at 50 A m−2 for an EC fed at the anode with synthetic urine, however, only 0.77 mol N L−1 d−1 was removed. Operation with real urine at 50 A m−2 achieved 63% removal with a competitive 0.59 mol N L−1 d−1. These data show the advantage of direct electrochemicalassisted stripping described in this paper on operational energy requirements. In a MEC fed with five times diluted real urine and operated at 14.6 ± 1.7 A m−2, nitrogen removal reached 34.2% (0.39 mol N L−1 d−1).23 The EC studied by Luther et al.21 resulted in 75 ± 0.5% (0.24 mol N L−1 d−1) nitrogen removal at 40 A m−2 with undiluted real urine. In addition, they evaluated NH3 stripping on hydrolyzed real urine in batch for 77 h at room temperature and removed 94% nitrogen (0.06 mol N L−1 d−1). Antonini et al.15 continuously stripped urine at pH 10 and 40 °C achieving 91% (1.4 mol N L−1 d−1) nitrogen removal at 5 h. Xu et al.19 reached 95% (0.16 mol N L−1 d−1) nitrogen removal by continuously stripping urine at pH 11 and 35 °C for 20−24 h. Thus, nitrogen removal rates and efficiencies reported here are the highest among (M)EC and reached similar values compared to alkaline stripping. Absolute TAN removal rates were only better for conventional stripping at pilot scale.15 Moreover, the studied EC configuration requires no heating or caustic addition. Relying on wind or solar energy this could result in a more sustainable technology compared to conventional stripping. Electromigration or Diffusion of NH4+ Depends on Concentration and (M)EC Configuration. In a (M)EC, cations can move through a CEM by diffusion following a concentration gradient or by electromigration, even against a concentration gradient.29 Ions present at high concentrations and with a high specific mobility, a.o. determined by the ion size and format, preferentially balance charge during electromigration, for example, NH4+ crossing a CEM toward the cathode rather than the smaller sized H+ in a neutral anolyte.22,30 Our EC with catholyte effluent as anolyte feed showed the highest concentration for Na+ followed by TAN, K+, and H+. Mainly Na+ and K+ balanced charge by electromigration because TAN diffused through the CEM as NH4+ due to NH3 removal from the catholyte. Protons were not abundant enough to represent a considerable fraction of electromigrating ions. In an MFC fed with urine in the anode compartment, NH 4 + electromigrated to the cathode. 24 Ammonium diffusion over the CEM also occurred because NH4+ was immediately converted in NH3 at the alkaline air cathode resulting in a NH4+ concentration gradient. In an EC with urine in the anode compartment, NH4+ crossed the CEM via electromigration.21 However, the Coulombic efficiency for NH4+ was only 55% at 40 A m−2 indicating other ions, H+/ OH−, Na+, and K+, in order of importance, also accounted for the charge balance.21 Optimizing NH3 Recovery Results in Competitive Energy Investments. Nitrogen recovery was lower compared to nitrogen removal, especially if HOB medium was used as

Table 2. Batch Growth Performance Parameters of Hydrogen Oxidizing Bacteria in a Bubble Column Fed with NH4Cl or TAN, Recovered from Urine by Absorption in Hob Medium, Compared to a Sequencing Batch Reactor (SBR)9a bubble column (this paper) Removal Efficiency (%) H2 O2 CO2 TAN Productivity (g CDW L−1 d−1) YH2 (g CDW g COD-H2−1) YCO2 (g CDW-C g CO2−C−1)

NH4Cl

urine TAN

SBR

60 ± 22 69 ± 17 74 ± 29 59 1.72 0.13 0.10

58 ± 17 77 ± 16 72 ± 12 48 1.85 0.17 0.16

65 ± 4 nd nd 100 4.49 ± 1.08 0.16 ± 0.04 0.25 ± 0.06

a

Averages (±SD) were calculated over the complete test duration (nNH4Cl ≥ 6, nurineTAN ≥ 10) for the bubble column and over three different runs for the SBR. Biomass productivity and yields were based on maximal values. nd = not determined.

Matassa et al. (2016),9 who obtained 11.2 g CDW L−1 in 6 days (5.6 ± 0.9 g CDW d−1) at a similar volumetric nitrogen loading rate. Most likely, this was due to the two times lower H2 and O2 volumetric loading rates compared to Matassa et al.9 Increasing H2 and O2 supply would increase NH3 uptake and biomass concentration.

■

DISCUSSION Competitive Nitrogen Removal Efficiencies at Zero Chemical NH3 Stripping. Compared to earlier studies with E

DOI: 10.1021/acs.est.7b02819 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

the urine to the MP culture. Additional gas filtration could act as an extra physical barrier to safeguard the MP culture. The combined effect of catholyte pH above 10 and the resulting shift to NH3 can inactivate pathogenic coliforms and viruses.32−35 Alternatively, changing the stripping and absorption in the studied EC configuration by a NH3 gas-permeable nanofiltration membrane could serve as a physical barrier for biological and chemical contamination of a potentially pure MP culture. In this way, a safe feed or even food product will be ensured making the consumption of waste-derived products more acceptable to the public opinion. A local source-separated urine treatment plant could recover nitrogen and return the produced microbial protein to the local market. This would avoid the dependence on soy or fishmeal import. To identify the economic impact of urine nitrogen supply for HOB production, a preliminary calculation was conducted with volumetric productivities and gas supplies achieved in an SBR.9 HOB production (12% N) would yield 4.3 ton CDW d−1 if the optimized urine electrolysis and stripping technology would treat 1.5 L urine IE−1 d−1 of 100 000 IE at 5 g N L−1 of which 68.4% could be recovered at 2.6 kWh kg N−1. If no losses occur and one needs 56 g protein or 9 g N d−1,37,38 this 0.51 ton N d−1 could locally feed 56 666 IE while avoiding wastewater treatment costs. A HOB growth reactor of 2300 m3 would be required with an operational energy need of 1100 kWh or 4000 MJ per kg CDW-N (H2 and O2 electrolytically produced at 83% CE; excluding pump costs, and CO2 off-gas pretreatment). This resembles energy requirements of conventional meat protein production although at higher nitrogen usage efficiencies and far less land use.2 Mainly H2 accounts for this energy requirement, although solar or wind energy can make electrolytic H2 production sustainable. In this way, zero chemical upgrading of waste-derived NH3 and off-gas CO2 into autotrophic MP could be established.

absorbent (SI S7 Table). The HOB medium had a neutral pH and thus lower absorption capacity compared to the acid absorbent. Although medium residence times were lower compared to acid residence times (SI S8 Figure), the applied 14−32 h still caused ammonia breakthrough (final absorbent pH 9.8 ± 0.4). Losses probably also occurred due to NH3 solubilization in the condensate in the stripping and absorption columns and the water trap. Such aspects could be avoided by replacing the stripping and absorption column by a gaspermeable hydrophobic membrane unit as recently described,31 but optimizing this recovery approach was not within the scope of this study. Nitrogen removal and recovery comes with an energy input reported as primary energy (excluding pump operation).1 The highest energy needs were calculated for real urine with HOB medium as absorbent: 3.6 and 19.4 kWh kg N−1 for removal and recovery, respectively, at 20 A m−2 and 2.7 ± 0.1 V cell voltage (SI S7 Table). Taking into account the energy content of H2 (33.3 kWh kg−1) produced at 28.5 ± 16.1% CE and aeration energy savings via O2 production (2 kWh kg−1) at 61.2 ± 32.7% CE, energy needs lowered to 2.6 and 13.9 kWh kg N−1 for nitrogen removal and recovery, respectively. These are competitive values compared with other urine (M)ECs that achieved 8 to 13.7 kWh kg N−1 for TAN removal with acid absorbents.21,23 If NH3 recovery in HOB medium can be optimized, for example, by a membrane unit operated at low absorbent HRT, recovery energy needs will decrease. Nitrogen recovery for subsequent MP production can than provide a local alternative for the Haber-Bosch process (10.3−12.5 kWh kg N−1 produced) and conventional gas stripping (3.9 to 28.2 kWh kg N−1, scale- and operation-dependent) while delivering additional energy benefits such as H2 production and saving the nitrogen removal costs (4.4−11.1 kWh kg N−1) in case the urine ends up in wastewater treatment plants.1 Environmental costs for reactive nitrogen loss could be avoided. Moreover, the proposed reactor concept can be fully driven by electricity, implying low-CO2 emitting power sources can be used. Urine Electrolysis Enables Microbial Protein Production. Besides the urine electrochemical cell for NH3 supply, an additional electrolyzer for H2 and O2 running on renewable energy and off-gas as a CO2 source are required for microbial protein production as HOB. External electrolysis is the more energy efficient option avoiding Cl2 generation from urine at an increasingly acidifying anode with current density. Experiments described here were run with excess nitrogen and limiting H2 and O2 supply. Compared to Matassa et al.,9 this was equally efficient based on gas removal efficiencies despite lower nitrogen removal efficiencies and biomass yields. Excess supply of H2 and O2 would favor nitrogen uptake and thus protein production. In terms of nitrogen, ammonia recovery from urine would be the process limiting step, not NH3 to MP conversion. Similar HOB biomass productivities on NH3 and NH4+ were observed. Ammonia, however, had two major advantages. First, a HOB reactor acidifies requiring caustic addition, which comes with additional salt stress and chemical costs. Ammonia from urine dissolved in the HOB medium, consuming H+ to form NH4+ and generating OH−. Feeding NH3 rather than NH4+ thus leads to improved buffering counteracting the acidifying nature of a HOB reactor and thus avoiding extra alkaline salt dosing. Consequently, low-salinity MP could find its way to the feed market. Second, NH3 was extracted from urine passing through the gas phase in a stripping column. This could eliminate potential pathogen and micropollutant transfer from

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.est.7b02819. Additional information as noted in the text (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +32 (0)9 264 59 76; fax: +32 (0)9 264 62 48; e-mail: [email protected]. ORCID

Marlies E.R. Christiaens: 0000-0002-2244-8952 Korneel Rabaey: 0000-0001-8738-7778 Present Addresses †

Organic Waste Systems (OWS), Dok Noord 5, B-9000 Gent, Belgium. ‡ AVECOM, Industrieweg 122P, B-9032 Wondelgem, Belgium. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS In this work, M.C. was supported by Ghent University and ESA/BELSPO for MELiSSA, S.G. by the Special Research Fund (BOF) from Ghent University, S.M. by Marie Curie MERMAID-ITN (FP7 European Commission, Grant No. F

DOI: 10.1021/acs.est.7b02819 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

(10) Yu, J.; Dow, A.; Pingali, S. The energy efficiency of carbon dioxide fixation by a hydrogen-oxidizing bacterium. Int. J. Hydrogen Energy 2013, 38 (21), 8683−8690. (11) Rose, C.; Parker, A.; Jefferson, B.; Cartmell, E. The characterization of feces and urine: A review of the literature to inform advanced treatment technology. Crit. Rev. Environ. Sci. Technol. 2015, 45 (17), 1827−1879. (12) Seneca, H.; Nally, R.; Peer, P. Microbial urease. Nature 1962, 193 (4820), 1106−1107. (13) Mobley, H. L. T.; Hausinger, R. P. Microbial ureases significance, regulation, and molecular characterization. Microbiological Reviews 1989, 53 (1), 85−108. (14) Basakcilardan-Kabakci, S.; Ipekoglu, A. N.; Talini, I. Recovery of ammonia from human urine by stripping and absorption. Environ. Eng. Sci. 2007, 24 (5), 615−624. (15) Antonini, S.; Paris, S.; Eichert, T.; Clemens, J. Nitrogen and phosphorus recovery from human urine by struvite precipitation and air stripping in vietnam. Clean: Soil, Air, Water 2011, 39 (12), 1099− 1104. (16) Morales, N.; Boehler, M. A.; Buettner, S.; Liebi, C.; Siegrist, H. Recovery of n and p from urine by struvite precipitation followed by combined stripping with digester sludge liquid at full scale. Water 2013, 5 (3), 1262−1278. (17) Böhler, M. Ammoniakstrippung mittels luft zur behandlung von faulwasser und urin auf der kläranlage kloten/opfikon (Air stripping of ammonia for the treatment of digester supernatant and urine at the wwtp kloten/opfikon). Aqua Gas 2012, 92 (1), 26−31. (18) Gulyas, H.; Zhang, S.; Otterpohl, R. Pretreating stored human urine for solar evaporation by low-technology ammonia stripping. J. Environ. Prot. 2014, 5, 962−969. (19) Xu, K.; Zhang, C.; Li, J.; Cheng, X.; Wang, C. Removal and recovery of n, p and k from urine via ammonia stripping and precipitations of struvite and struvite-k. Water Sci. Technol. 2017, 75 (1), 155−164. (20) Gildemyn, S.; Luther, A. K.; Andersen, S. J.; Desloover, J.; Rabaey, K. Electrochemically and bioelectrochemically induced ammonium recovery. J. Visualized Exp. 2015, No. 95, e52405. (21) Luther, A. K.; Desloover, J.; Fennell, D. E.; Rabaey, K. Electrochemically driven extraction and recovery of ammonia from human urine. Water Res. 2015, 87, 367−377. (22) Desloover, J.; Woldeyohannis, A. A.; Verstraete, W.; Boon, N.; Rabaey, K. Electrochemical resource recovery from digestate to prevent ammonia toxicity during anaerobic digestion. Environ. Sci. Technol. 2012, 46 (21), 12209−12216. (23) Kuntke, P.; Sleutels, T.; Saakes, M.; Buisman, C. J. N. Hydrogen production and ammonium recovery from urine by a microbial electrolysis cell. Int. J. Hydrogen Energy 2014, 39 (10), 4771−4778. (24) Kuntke, P.; Smiech, K. M.; Bruning, H.; Zeeman, G.; Saakes, M.; Sleutels, T.; Hamelers, H. V. M.; Buisman, C. J. N. Ammonium recovery and energy production from urine by a microbial fuel cell. Water Res. 2012, 46 (8), 2627−2636. (25) Bonvin, C.; Etter, B.; Udert, K. M.; Frossard, E.; Nanzer, S.; Tamburini, F.; Oberson, A. Plant uptake of phosphorus and nitrogen recycled from synthetic source-separated urine. Ambio 2015, 44, S217−S227. (26) Greenberg, A. E. Standard Methods for the Examination of Water and Wastewater; American Public Health Association: Washington DC, 1992. (27) Vilchez-Vargas, R.; Geffers, R.; Suarez-Diez, M.; Conte, I.; Waliczek, A.; Kaser, V. S.; Kralova, M.; Junca, H.; Pieper, D. H. Analysis of the microbial gene landscape and transcriptome for aromatic pollutants and alkane degradation using a novel internally calibrated microarray system. Environ. Microbiol. 2013, 15 (4), 1016− 1039. (28) Marchesi, J. R.; Sato, T.; Weightman, A. J.; Martin, T. A.; Fry, J. C.; Hiom, S. J.; Wade, W. G. Design and evaluation of useful bacterium-specific pcr primers that amplify genes coding for bacterial 16s rrna. Appl. Environ. Microbiol. 1998, 64 (2), 795−799.

607492), J.D. as postdoctoral fellow by Research Foundation Flanders (FWO), and K.R. by a European Research Council Starter Grant ELECTROTALK. All authors acknowledge the support by BOF Basisinfrastructuur (Grant No. 01B05912) for equipment used in this study. We acknowledge Robin Declerck for technical assistance and thank Ramon Ganigué, Amanda Luther, and Jan B.A. Arends for critically reading and commenting on the manuscript.

■

ABBREVIATIONS MP microbial protein HOB hydrogen oxidizing bacteria N nitrogen TAN total ammonium nitrogen TKN total kjeldahl nitrogen (M)EC microbial electrochemical cell MFC microbial fuel cell j current density CE Coulombic efficiency CEM cation exchange membrane HRT hydraulic retention time Q flow rate SD standard deviation SBR sequencing batch reactor IE inhabitant equivalant IC ion chromatography TSS total suspended solids CDW cell dry weight tCOD total chemical oxygen demand sCOD soluble chemical oxygen demand cBOD5 carbonaceous biological oxygen demand over 5 days VFA volatile fatty acids PCR polymerase chain reaction DNA deoxyribonucleic acid

■

REFERENCES

(1) Maurer, M.; Schwegler, P.; Larsen, T. A. Nutrients in urine: Energetic aspects of removal and recovery. Water Sci. Technol. 2003, 48 (1), 37−46. (2) Matassa, S.; Batstone, D. J.; Hülsen, T.; Schnoor, J.; Verstraete, W. Can direct conversion of used nitrogen to new feed and protein help feed the world? Environ. Sci. Technol. 2015, 49 (9), 5247−5254. (3) McAllister, C. H.; Beatty, P. H.; Good, A. G. Engineering nitrogen use efficient crop plants: The current status. Plant Biotechnology Journal 2012, 10 (9), 1011−1025. (4) Vitousek, P. M.; Mooney, H. A.; Lubchenco, J.; Melillo, J. M. Human domination of earth’s ecosystems. Science 1997, 277 (5325), 494−499. (5) Pan, B. B.; Lam, S. K.; Mosier, A.; Luo, Y. Q.; Chen, D. L. Ammonia volatilization from synthetic fertilizers and its mitigation strategies: A global synthesis. Agric., Ecosyst. Environ. 2016, 232, 283− 289. (6) Anupama; Ravindra, P. Value-added food:: Single cell protein. Biotechnol. Adv. 2000, 18 (6), 459−479. (7) Puyol, D.; Batstone, D.; Hülsen, T.; Astals, S.; Peces, M.; Krömer, J. Resource recovery from wastewater by biological technologies: Opportunities, challenges and prospects. Front. Microbiol. 2016, 7 (2106), 1−23. (8) Goldberg, I. Single cell protein; Springer: Berlin Germany, 1985. (9) Matassa, S.; Verstraete, W.; Pikaar, I.; Boon, N. Autotrophic nitrogen assimilation and carbon capture for microbial protein production by a novel enrichment of hydrogen-oxidizing bacteria. Water Res. 2016, 101, 137−146. G

DOI: 10.1021/acs.est.7b02819 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology (29) Cheng, K. Y.; Kaksonen, A. H.; Cord-Ruwisch, R. Ammonia recycling enables sustainable operation of bioelectrochemical systems. Bioresour. Technol. 2013, 143, 25−31. (30) Cord-Ruwisch, R.; Law, Y.; Cheng, K. Y. Ammonium as a sustainable proton shuttle in bioelectrochemical systems. Bioresour. Technol. 2011, 102 (20), 9691−9696. (31) Sleutels, T.; Hoogland, B. J.; Kuntke, P.; ter Heijne, A.; Buisman, C. J. N.; Hamelers, H. V. M. Gas-permeable hydrophobic membranes enable transport of CO2 and NH3 to improve performance of bioelectrochemical systems. Environmental Science-Water Research & Technology 2016, 2 (4), 743−748. (32) Allievi, L.; Colombi, A.; Calcaterra, E.; Ferrari, A. Inactivation of fecal bacteria in sewage-sludge by alkaline treatment. Bioresour. Technol. 1994, 49 (1), 25−30. (33) Decrey, L.; Kazama, S.; Udert, K. M.; Kohn, T. Ammonia as an in situ sanitizer: Inactivation kinetics and mechanisms of the ssrna virus ms2 by nh3. Environ. Sci. Technol. 2015, 49 (2), 1060−1067. (34) Gajda, I.; Greenman, J.; Melhuish, C.; Ieropoulos, I. A. Electricity and disinfectant production from wastewater: Microbial fuel cell as a self-powered electrolyser. Sci. Rep. 2016, 6, 1−9. (35) Mendez, J. M.; Jimenez, B. E.; Barrios, J. A. Improved alkaline stabilization of municipal wastewater sludge. Water Sci. Technol. 2002, 46 (10), 139−146. (36) Arrendondo, M. R.; Kuntke, P.; ter Heijne, A.; Hamelers, H. V. M.; Buisman, C. J. N. Load ratio determines the ammonia recovery and energy input of an electrochemical system. Water Res. 2017, 111, 330−337. (37) Pendick, P. How much protein do you need every day? Harvard Health Blog 2015, http://www.health.harvard.edu/blog/how-muchprotein-do-you-need-every-day-201506188096 (accessed October 10, 2017). (38) Mariotti, F.; Tomé, D.; Patureau Mirand, P. Converting Nitrogen into ProteinBeyond 6.25 and Jones’ Factors. Crit. Rev. Food Sci. Nutr. 2008, 48, 177−184.

H

DOI: 10.1021/acs.est.7b02819 Environ. Sci. Technol. XXXX, XXX, XXX−XXX