view article online envonir ment l a...

This is an Accepted Manuscript, which has been through the Royal Society of Chemistry peer review process and has been accepted for publication.

Accepted Manuscripts are published online shortly after acceptance, before technical editing, formatting and proof reading. Using this free service, authors can make their results available to the community, in citable form, before we publish the edited article. We will replace this Accepted Manuscript with the edited and formatted Advance Article as soon as it is available.

You can find more information about Accepted Manuscripts in the Information for Authors.

Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content. The journal’s standard Terms & Conditions and the Ethical guidelines still apply. In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains.

Accepted Manuscript

rsc.li/es-water

Environmental Science Water Research & Technology

View Article OnlineView Journal

This article can be cited before page numbers have been issued, to do this please use: V. Srinivasan, J.

Weinrich and C. Butler, Environ. Sci.: Water Res. Technol., 2016, DOI: 10.1039/C5EW00260E.

http://www.rsc.org/Publishing/Journals/guidelines/AuthorGuidelines/JournalPolicy/accepted_manuscripts.asp

http://www.rsc.org/help/termsconditions.asp

http://www.rsc.org/publishing/journals/guidelines/

http://dx.doi.org/10.1039/c5ew00260e

http://pubs.rsc.org/en/journals/journal/EW

http://crossmark.crossref.org/dialog/?doi=10.1039/C5EW00260E&domain=pdf&date_stamp=2016-01-11

This study presents the conditions of nitrite accumulation in MFC biocathodes through batch

experiments and derives kinetic parameters with an Activated Sludge Model with an integration of the

Nernst-Monod model and Indirect Coupling of Electrons (ASM-NICE).

Page 1 of 11 Environmental Science: Water Research & Technology

Env

iron

men

talS

cien

ce:W

ater

Res

earc

h&

Tech

nolo

gyA

ccep

ted

Man

uscr

ipt

Publ

ishe

d on

11

Janu

ary

2016

. Dow

nloa

ded

on 1

2/01

/201

6 12

:57:

13.

View Article OnlineDOI: 10.1039/C5EW00260E



PAPER

This journal is © The Royal Society of Chemistry 20xx J. Name., 2013, 00, 1-3 | 1

Please do not adjust margins


Received 00th January 20xx, 1 Accepted 00th January 20xx 2

DOI: 10.1039/x0xx00000x 3

www.rsc.org/ 4

5

Nitrite Accumulation in a Denitrifying Biocathode Microbial Fuel 6

Cell 7

Varun Srinivasana, Jacob Weinricha, Caitlyn Butlera,† 8

Microbial fuel cells (MFCs) are a potential treatment technology-energy requirements for treatment can be offset with 9 electricity production, biomass yield can be minimized, and microbial electron donors can be decoupled from acceptors, 10 expanding treatment options. One potential MFC configuration uses an organic-oxidizing anode biofilm and a denitrifying 11 cathode biofilm. However nitrite, a denitrification intermediate with environmental and public health impacts, has been 12 reported to accumulate. In this study, before complete denitrification was achieved in a bench-scale, batch denitrifying 13 cathode, nitrite concentrations reached 66.4 % ± 7.5 % of the initial nitrogen. Common environmental inhibitors such as 14 insufficient electron donor, dissolved oxygen, insufficient carbon source, and pH, were considered. Improvement in these 15 conditions did not mitigate nitrite accumulation. We present an Activated Sludge Model with an integration of the Nernst-16 Monod model and Indirect Coupling of Electrons (ASM-NICE) that effectively simulated the observed batch data, including 17 nitrite-accumulation by coupling biocathodic electron transfer to intracellular electron mediators. The simulated half-18 saturation constants for mediated intracellular transfer of electrons during nitrate and nitrite reduction suggested a greater 19 affinity for nitrate reduction when electrons are not limiting. The results imply that longer hydraulic retention times (HRTs) 20 may be necessary for a denitrifying biocathode to ensure complete denitrification. These findings could play a role in 21 designing full-scale MFC wastewater treatment systems to maximize total nitrogen removal. 22

Water Impact 23

Organics-oxidizing microbial fuel cells (MFCs) with a denitrifying cathode are a potential wastewater treatment technology that could offset energy 24 requirements and minimize biomass yield. To improve the efficacy of this approach, we investigated the accumulation of nitrite during denitrification and 25 modeled biocathodic denitrification processes using an Activated Sludge Model with an integration of the Nernst-Monod model and Indirect Coupling of 26 Electrons (ASM-NICE). 27

Introduction 28

Microbial fuel cells (MFC) have emerged as a potentially 29 energy-efficient treatment strategy with promising 30 applications in wastewater1–4, in-situ environmental 31 remediation5–7 and decentralized treatment systems8. A 32 particular advantage of MFC application as a treatment 33 strategy is the ability to decouple the electron donor from 34 the electron acceptor in biological reactions, allowing the 35 oxidation of natural or wastewater organics to facilitate the 36 reduction of an oxidized contaminant at the cathode. A 37 variety of oxidized contaminants have been reduced in this 38 way including nitrate, perchlorate, uranium, trichloroethane 39

and dichromate5,7,9–13. Additionally, the electrons passed 40 across an external load between the anode and cathode can 41 offset energy requirements for treatment. 42

Nitrate is a contaminant of interest for drinking water 43 systems with a minimum contaminant level (MCL) of 10 mg 44 NO3-N /L in the United States14. Nitrate has also been 45 regulated in wastewater effluents through total nitrogen 46 discharge limits in an effort to curb eutrophication of surface 47 waters and other environmental impacts. Biological 48 nitrification-denitrification is one of the most common 49 processes used for total nitrogen removal from 50 wastewater15. Nitrifying bacteria oxidize ammonia to nitrite 51 and, then, nitrite to nitrate. Biological denitrification reduces 52 nitrate to nitrogen gas. Denitrification is a dissimilatory 53 microbiological process, which many heterotrophic and 54 autotrophic organisms are capable of performing. It is a 55 sequential reaction involving the reduction of nitrate to 56 nitrite by a nitrate reductase enzyme. Nitrite is reduced to 57 nitric oxide by a nitrite reductase enzyme. Nitric oxide is 58 reduced to nitrous oxide by a nitric oxide reductase enzyme 59

a. Department of Civil and Environmental Engineering, University of Massachusetts-Amherst, Amherst MA 01003.

†Corresponding Author: 18 Marston Hall, 130 Natural Resources Road, University of Massachusetts-Amherst, Department of Civil and Environmental Engineering, Amherst MA 01003, USA. Tel: (413)545-5396. Email: [email protected] b. Electronic Supplementary Information (ESI) available: See

DOI: 10.1039/x0xx00000x

Page 2 of 11Environmental Science: Water Research & Technology

Env

iron

men

talS

cien

ce:W

ater

Res

earc

h&

Tech

nolo

gyA

ccep

ted

Man

uscr

ipt

Publ

ishe

d on

11

Janu

ary

2016

. Dow

nloa

ded

on 1

2/01

/201

6 12

:57:

13.



ARTICLE Journal Name

2 | J. Name., 2012, 00, 1-3 This journal is © The Royal Society of Chemistry 20xx



after which nitrous oxide is reduced to nitrogen gas by a 60 nitrous oxide reductase16. 61

Though several studies have reported significant 62 denitrification in microbial fuel cells 10,11,17, some studies 63 have reported accumulation of nitrite of up to 50-55% of the 64 initial total nitrogen during autotrophic denitrification in 65 cathodes18,19 and accumulation of nitrous oxide of up to 70% 66 of the initial total nitrogen added as nitrate20. Minimizing the 67 accumulation of intermediates is critical to achieving 68 maximum nitrogen removal. Denitrification intermediates 69 have negative health and environmental effects. Nitrite can 70 cause methemoglobinemia in aquatic life, small children and 71 the elderly 21,22 and nitrous oxide is a potent greenhouse 72 gas23. 73

The accumulation of denitrification intermediates can 74 occur due to differing metabolic rates of the denitrification 75 steps. Additionally, several environmental factors could 76 contribute to this accumulation. Denitrification enzyme 77 production is repressed at DO concentrations above 2.5 mg-78 O2/L.16,24 Incomplete denitrification can be due to limiting 79 electron donor concentrations.16 Inhibition of denitrification 80 can occur outside the pH range of 7 to 8 leading to an 81 accumulation of intermediates. pH below 7 can cause direct 82 inhibition through the formation of free nitrous acid which is 83 a protonated form of nitrite.25 A pH above 8 can cause 84 inhibition of general microbiological processes. 85

Mathematical modeling has played a very critical role in 86 predicting nitrogen removal in wastewater treatment. The 87 modeling of accumulation of intermediates has been 88 achieved by modeling denitrification as a four-step 89 denitrification process, using reaction-specific kinetic rate 90 equations for each step. Broadly, two major models have 91 been proposed in recent years: the “direct-coupling” 92 approach used in Activated Sludge Model for Nitrogen 93 (ASMN) 26 and the “indirect-coupling” approach used in the 94 Activated Sludge Model for Indirect Coupling of Electrons 95 (ASM-ICE)27. The ASMN model directly couples the carbon 96 oxidation with each of the nitrogen oxide reduction steps. 97 The ASM-ICE model indirectly couples carbon oxidation and 98 nitrogen oxide reduction using intermediate electron 99 carriers. A comparison of these models in predicting 100 accumulation of nitrogen intermediates was made by Pan et 101 al. 28, concluding that the ASM-ICE model predicts the 102 accumulation of nitrogen oxides better than the ASMN 103 model. 104 The ASM-ICE model was formulated, assuming a dissolved 105 carbon/electron source and using the dual-substrate 106 limitation kinetics derived from the Monod model. The 107 Monod model works well for dissolved substrates but in a 108 denitrifying biocathode, the biofilm is limited by a solid 109 electron donor. In this case, the ability of the cathode 110 electrode to act as an electron donor is determined by the 111 cathode potential according to the following equation, which 112

has been adapted from a model prepared for a MFC anode 113 by Marcus et al.29 114 115

µ = − µ𝑚𝑎𝑥 𝑋𝑓𝐿𝑓(𝑆𝑎

𝐾𝑠𝑎+𝑆𝑎)(

1

1+exp[−𝐹

𝑅𝑇𝜂]

) (1) 116

where F is the Faraday’s constant, R is the gas constant, T is 117 the temperature (°C), Xf is the biofilm cell density and Lf is 118 the depth of the biofilm. and 𝜂 = (Ecathode- EKc) , where Ecathode 119 is the cathode potential and EKc is the cathodic mid-point 120 electron donor potential for the half-maximum rate 29. The 121 Nernst-Monod model for a biocathode can simulate electron 122 transfer from a cathode to associated microorganisms 123 assumed to be in a biofilm on the cathode surface.30,31 The 124 ASM-ICE model can be used to simulate nitrite accumulation 125 in denitrifying bioreactors.27 The incorporation of 126 bioelectrochemical elements into an ASM model has not 127 been previously described. An integration of the two models 128 could help elucidate performance parameters in BESs 129 designed for wastewater treatment. 130

Though complete denitrification has been demonstrated 131 in MFCs using a cathode as an electron donor for autotrophic 132 denitrification, more research is needed to understand the 133 conditions that facilitate incomplete denitrification and the 134 accumulation of intermediates. Additionally, the Monod 135 kinetic parameters, that are crucial to the design of 136 treatment processes, have not been determined for 137 autotrophic denitrification in a MFC cathode. In this study, 138 we investigated the mechanisms behind the accumulation of 139 nitrite by performing experiments under different 140 environmental conditions. We present an ASM with an 141 integration of the Nernst-Monod model and Indirect 142 Coupling of Electrons (ASM-NICE), for the simulation of 143 nitrogen removal in a denitrifying biocathode. We used this 144 model to estimate kinetic parameters for a denitrifying 145 biocathode. 146

Materials and Methods 147

MFC Configuration and Operation 148

Duplicate flat-plate MFCs were constructed from rectangular 149 Plexiglas frames (10 x 10 x 1.2 cm) and filled with graphite 150 granules (porosity=0.55). The total volume and liquid volume 151 of the electrode compartments were 120 mL and 54 mL 152 respectively. Graphite rods were inserted into the anode and 153 cathode chambers to act as electron collectors. The 154 compartments were separated by a cation exchange 155 membrane (CMI-7000, Membrane International, Glen Rock, 156 NJ) which was activated using a 5% NaCl solution at 40 C for 157 24 hours. Ag/AgCl reference electrodes (RE-6, BASi Inc. USA) 158 were used to monitor the cathode potentials. 159

The MFC anodes were inoculated with a combination of 160 anode effluent from a parent MFC (that was constructed and 161 operated similarly to the experimental MFCs) and primary 162


Env

iron

men

talS

cien

ce:W

ater

Res

earc

h&

Tech

nolo

gyA

ccep

ted

Man

uscr

ipt

Publ

ishe

d on

11

Janu

ary

2016

. Dow

nloa

ded

on 1

2/01

/201

6 12

:57:

13.



Journal Name ARTICLE




Table 1: Experimental Design 163

164 effluent from the Amherst Wastewater Treatment Plant 165 (WWTP), Amherst, MA. The cathodes were inoculated with a 166 combination of cathode effluent from the denitrifying 167 biocathode of the parent MFC, primary effluent from the 168 Amherst WWTP and a pond sediment inoculum from the 169 Campus Pond, University of Massachusetts-Amherst, MA. 170 The reactors were initially operated in batch mode where the 171 effluent of each electrode compartment was returned to the 172 influent to allow the accumulation of biofilms on the anode 173 and cathode. The end of a batch cycle was indicated by the 174 voltage decreasing to less than 0.05 V measured across the 175 anode and cathode. After 50 batch cycles, the anode 176 chamber was switched to a continuous flow mode. The 177 anode was operated in continuous flow to prevent electron 178 donor limiting conditions from the anode. The feed for the 179 anode was supplied at a flow rate of 0.25 mL/min (hydraulic 180 retention time=20.5 hours) resulting in a COD loading rate of 181 154 mg COD/L-day. The anode media during the course of 182 experiments, unless otherwise stated, consisted of (per liter) 183 1.386 g Na2HPO4, 0.849 g KH2PO4, 0.05 g NH4Cl, 0.05 g MgCl2, 184 and 0.710 g CH3COOK (0.355 g COD). The cathode remained 185 in batch operation during all experiments. In this 186 configuration, a single batch cycle typically lasted three days 187 and was marked by the voltage decreasing to less than 0.05 188 V measured across the anode and cathode. To promote 189 complete mixing, the cathode feed was recycled at 30 190 mL/min between the cathode and a 1 L external sealed 191

recycle bottle used for collection of gas and liquid samples. 192 The cathode media during the course of experiments, unless 193 otherwise stated, consisted of (per liter) 0.7098 g Na2HPO4, 194 1.4968 g KH2PO4, 0.05 g MgSO4, and 0.1228 g NaNO3. 195 Additionally, trace minerals were added to each solution, 196 including (per liter):1 mg CaCl2•2H2O, 1 mg FeSO4•7H2O, 100 197 µg ZnSO4•7H2O, 30 µg MnCl2•4H2O, 300 µg H3BO3, 200 µg 198 CoCl2•6H2O, 10 µg CuCl2•2H2O, 10 µg NiCl2•6H2O, 30 µg 199 Na2MoO4•2H2O, and 30 µg Na2SeO3. All the feed solutions 200 were sparged with N2 before the start of each batch cycle. 201 The MFCs and feed bottles were covered with aluminium foil 202 to ensure light did not inhibit denitrification or cause 203 phototrophic growth. After the acclimation period and 204 establishing steady state conditions, experiments were 205 performed to obtain data for model fitting for the purpose of 206 estimating denitrification kinetic parameters (Table 1, E2 207 &E3). Experiments were also performed to determine if 208 environmental parameters such as dissolved oxygen 209 concentration, pH and carbon limitation were responsible 210 for the nitrite accumulation (Table 1, E4-E9). 211 212

Analyses and Calculations 213

Nitrate, nitrite, acetate, and sulfate were monitored in 214 the influent and effluent of the anode and cathode 215 compartments using a Metrohm 850 Professional Ion 216 Chromatograph (IC) (Metrohm Inc., Switzerland) with a 217 Metrosep A Supp 5-250 3.2 mM Na2CO3, 1.0 mM NaHCO3 218 eluent was pumped at 2.6 mL/min, with a 100 mM HNO3 219 suppressor solution and using a 20 µL sample loop. 220 Ammonium was monitored using a Metrohm 850 221 Professional IC (Metrohm Inc., Switzerland) with a Metrosep 222 C 2-250 cation column (Metrohm Inc., Switzerland). For 223 cation analysis, an eluent consisting of 0.75 mM dipicolinic 224 acid and 4 mM tartaric acid was pumped at 1 mL/min, using 225 a 10 µL sample loop. Each sample was filtered through 0.1 226 µm syringe filters, stored at 4 C and analyzed within 5 days 227 of sampling. 228

Nitric oxide and nitrous oxide were measured using an 229 Agilent 7890A Gas Chromatograph with a Thermal 230 Conductivity Detector (Agilent, USA) with HP-PLOT 231 Molesieve column (Agilent, USA). The inlet was heated to a 232 temperature of 200°C, with a total flow of 35.5 mL/min and 233 the septum flow at 3 mL/min. A split inlet was used at a 4:1 234 ratio, or at 26 mL/min. The initial oven temperature at the 235 beginning of each run was 35°C and held for 5 minutes, then 236 ramped at 25°C/min to 200°C, where it was held for 4 237 minutes. The TCD filament was set at 250°C, with a 20 238 mL/min reference flow, and a 4.5 mL/min makeup flow. The 239 makeup gas used was helium. Inorganic carbon was 240 measured using a Shimadzu TOC-VCPH Total Organic Carbon 241 Analyzer (Shimadzu, Japan). pH was measured using a Fisher 242 Science Education pH Meter (Fisher Scientific, USA). 243 244 Electrochemical Analyses 245

Experiment Media Change Purpose

E1 No Change Acclimation of

biofilm to reach

steady state

conditions

E2 No Change Data collection for

model calibration

E3 20 mg NO3--N/L replaced with

30 mg NO2--N/L

Data collection for

nitrite kinetic

parameters

E4 Acetate concentration in the

anode increased from 154

mg-COD/L to 770 mg-COD/L

Determine if electron

donor was limiting

E5 Catholyte constantly sparged

with N2 in Recycle Bottle

during Cycle

Eliminate potential

DO diffusion in

cathode

E6 Catholyte amended with 5

mg HCO3--C/L

Determine if

inorganic carbon is

limiting

E7 30 mg NO2--N/L, starting pH

6.6

Determine if pH

lowers nitrite

reduction rate


7.0

Determine if pH

lowers nitrite

reduction rate


7.4

Determine if pH

lowers nitrite

reduction rate


Env

iron

men

talS

cien

ce:W

ater

Res

earc

h&

Tech

nolo

gyA

ccep

ted

Man

uscr

ipt

Publ

ishe

d on

11

Janu

ary

2016

. Dow

nloa

ded

on 1

2/01

/201

6 12

:57:

13.







Table 2: Process Matrix for ASM-NICE 246

Process SNA SNI SMox SMred Rate Expression

R1 -0.5 0.5

𝑗 = −𝑗𝑚𝑎𝑥(1

1 + exp (−𝐹𝜂𝑅𝑇

))(

𝑆𝑀𝑜𝑥

𝑆𝑀𝑜𝑥 + 𝐾𝑀𝑜𝑥)

R2 -1 1 1 -1

𝑅𝑁𝐴 = 𝑟𝑁𝐴_𝑚𝑎𝑥𝑋𝑓(𝑆𝑀𝑟𝑒𝑑

𝑆𝑀𝑟𝑒𝑑 + 𝐾𝑀𝑟𝑒𝑑𝑁𝐴)(

𝑆𝑁𝐴

𝑆𝑁𝐴 + 𝐾𝑁𝐴)

R3 -1 +0.5 -0.5

𝑅𝑁𝐼 = 𝑟𝑁𝐼_𝑚𝑎𝑥𝑋𝑓(𝑆𝑀𝑟𝑒𝑑

𝑆𝑀𝑟𝑒𝑑 + 𝐾𝑀𝑟𝑒𝑑𝑁𝐼)(

𝑆𝑁𝐼

𝑆𝑁𝐼 + 𝐾𝑁𝐼)

R4 𝐶𝑡𝑜𝑡 = 𝑆𝑀𝑜𝑥 + 𝑆𝑀𝑟𝑒𝑑

247

The MFCs were operated with a 100-Ω external resistance. 248 The potential difference across the external resistances was 249 monitored every 10 minutes using a Keithley Model 2700 250 Multimeter with a 7700 Switching Module (Keithley 251 Instruments Inc., Cleveland, OH, USA). Low scan rate cyclic 252 voltammetry (LSCV) were performed using a Gamry Series 253 G750 Potentiostat/Galvanostat/ZRA (Gamry, USA). The 254 cathode potential was swept from -0.4 V vs SHE to 0.4 V vs 255 SHE at 1 mV/s and the current density was recorded. jmax and 256 EKc were estimated by fitting the Nernst-Monod equation to 257 the LSCV curve.32 258 259 Theory and Modeling 260

The model presented in this study was obtained through an 261 integration of the Nernst-Monod model29 with the ASM-ICE 262 model.27 The rate equations and the process matrix are 263 presented in Table 2. Briefly, the cathode oxidation and 264 simultaneous reduction of the oxidized intracellular electron 265 carrier (SMox) is represented by R1 which simulates electron 266 transfer from the cathode to the biofilm. Nitrate and nitrite 267 reduction are represented by R2 and R3 respectively. The 268 transfer of electrons from cathode oxidation to the 269 reduction of nitrate and nitrite is accomplished in the model 270 through the reduction of the SMox to SMred, which is 271 mathematically represented through a constant total 272 concentration (Ctot) as R4. The parameters represented in 273 the model are as follows: RNA is the nitrate reduction rate 274 (mmol/(g-VSS.L.h)), RNI is the nitrite reduction rate 275 (mmol/(g-VSS.L.h)), SNA and SNI are the nitrate and nitrite 276 concentrations respectively (mmol/L), j is the current 277 density (A/m2), η=Ecat-EKC where Ecat is the cathode potential 278 and EKC is the cathodic electron donor potential for the half-279 maximum rate (V), jmax is the maximum current density 280 (A/m2), SMox and SMred are the concentrations (mmol/gVSS) 281 of oxidized and the reduced form of the intermediate 282 electron carrier, Ctot is the total concentration of the electron 283

carrier (mmol/gVSS), KMox is the half saturation constant for 284 SMox (mmol/gVSS), KMredNA is the SMred affinity constant for 285 nitrate reductase (mmol/gVSS) and KMredNI is the SMred 286 affinity constant for nitrite reductase (mmol/gVSS), rNA_max 287 and rNI_max are maximum nitrate and nitrite reduction rates 288 (mmol/(gVSS.h)) respectively. 289 All kinetics parameters were estimated by solving the 290 differential rate equations and fitting the modeled data to 291 observed data. The following assumptions were made for 292 the purpose of modeling: the biofilm on the cathode was at 293 steady state i.e. growth of the biofilms equals decay and 294 detachment and for thin biofilms such as those observed in 295 denitrifying cathodes, diffusional limitations are minimal. 296 This assumption was confirmed with biofilm imaging (data 297 not shown). The total electron carrier concentration, Ctot was 298 assumed to be 0.01 (mmol/(gVSS)) as in Pan et al. (2013).27 299 KMox was assumed to be 1% of the Ctot to ensure that the 300 reduction of SMox was not rate limiting. KNA was assumed to 301 be 3.21 X 10-3 mmol/L.33 The biomass parameter (XfLf) was 302 determined by dividing the amount of volatile suspended 303 solids (VSS) by the projected surface area of the cathode. 304 The amount of VSS was determined by using Standard 305 Methods 2540 D and 2540 E at the end of operation of the 306 MFCs. The projected surface area of the cathode electrode 307 was determined by performing a particle size distribution 308 analysis and determining an average particle size. The 309 assumption was made that the particles were spherical. The 310 d50 or 50 % particle size was used as the average particle 311 size. The specific surface area (SSA) was calculated using the 312 following equation 313

𝑆𝑆𝐴 =6∗(1−𝜃)

𝑑50 (2) 314

where SSA is the specific surface area (m2/m3), θ is the 315 packed bed porosity and d50 is the 50% passing particle size. 316

All the modeling was done using R statistical software.34–317 36 Parametric estimations was done by minimizing the sum 318


Env

iron

men

talS

cien

ce:W

ater

Res

earc

h&

Tech

nolo

gyA

ccep

ted

Man

uscr

ipt

Publ

ishe

d on

11

Janu

ary

2016

. Dow

nloa

ded

on 1

2/01

/201

6 12

:57:

13.







319

Figure 1: A Typical Batch Cycle Observation during Denitrification. The lines are 320 there only for highlighting trends and are not representative of model simulations. 321

of squares of the residuals and using a genetic algorithm, 322 similar to the implementation by Pelletier et al.37 323 Model Calibration and Validation 324

The nitrite parameters (KNI, rNI_max) were estimated by 325 calibrating the model to the data for a representative batch 326 from experiment E3 where nitrite was added to the media 327 instead of nitrate. The rest of the parameters were 328 estimated by calibrating the model to a typical batch data 329 (Figure 1, replicate batches presented in Figure S1) using a 330 genetic algorithm with multi-objective optimization which 331 was implemented using a non-dominated sort method.38The 332 half-saturation constant for nitrate reduction (KNA) was 333 assumed from literature33, since nitrate reduction has been 334 extensively studied. The multi-objective optimization was 335 performed using nitrate and nitrite model fits as the two 336 objectives for data from E2. Model validation was performed 337 with two separate batches from experiment E1 from two 338 different reactors to test and demonstrate applicability of 339

340 341 Figure 2: Polarization curves performed during different stages of denitrification 342

the model to other reactors acclimated under the same 343 conditions. 344

Results and Discussion 345

Denitrifying Biocathode Acclimation 346

During the initial acclimation period (~50 batch cycles), the 347 denitrification rates gradually increased and became 348 consistent, indicating the establishment of a steady-state 349 biofilm. During a typical batch of the biocathode, the 350 removal of nitrate with simultaneous accumulation of nitrite 351 was observed (Figure 1A). Subsequent to the removal of 352 nitrate, removal of accumulated nitrite was observed. Peak 353 nitrite accumulation observed was 66.4 ± 7.5 % of the initial 354 nitrogen concentration added as nitrate. This is consistent 355 with previously reported nitrite peak accumulation values of 356 50-55 % of initial nitrogen added as nitrate 18,19. The 357 accumulation of nitrite was reproducible in different batches 358 in the duplicate reactors (Figure S1). Nitrous oxide also 359 accumulated in the recycle bottle headspace, with a peak 360 accumulation of 1% of the initial nitrogen added as nitrate 361 (Figure 1A). Previous studies have reported nitrous oxide 362 accumulation of ~0.025 % to 70 % of the initial nitrogen 363 added. 18,20,39 It should be noted that at the end of each 364 batch cycle denitrification was complete and the average 365 denitrification rate was 5.8 ± 0.4 g-N/(m3-d). The cathode 366 potential during the various stages of denitrification was also 367 monitored. It decreased from -0.014 ±0.007 V vs SHE during 368 nitrate reduction to -0.038 ± 0.004 V vs SHE during nitrite 369 reduction. Once nitrite was depleted, the cathode potential 370 dropped to – 0.254 ± 0.007 V vs SHE (Figure 1B). 371

Volumetric power density (W/m3) also followed a similar 372 trend decreasing during the difference stages of the 373 denitrification. Polarization curves were performed at 374 different stages of a single batch cycle. A curve was obtained 375 when nitrate was the primary dissolved form of nitrogen and 376


Env

iron

men

talS

cien

ce:W

ater

Res

earc

h&

Tech

nolo

gyA

ccep

ted

Man

uscr

ipt

Publ

ishe

d on

11

Janu

ary

2016

. Dow

nloa

ded

on 1

2/01

/201

6 12

:57:

13.







primary available acceptor (at 5 hours), when nitrite was the 377 primary electron acceptor available (at 45 hours) and when 378 nitrous oxide was the remaining available electron acceptor 379 (at 70 hours) (Figure 2). The maximum power production 380 during nitrite reduction (0.97 ± 0,21 W/m3-total cathode 381 volume (TCV)) only achieved 64% of the maximum power 382 produced during nitrate reduction (1.51±0.29 W/m3-TCV). 383 Very little power production was observed during nitrous 384 oxide reduction (0.03 ± 0.005 W/m3-TCV). This suggests that 385 the predominant reduction pathway in the cathode can 386 dictate the power production in an MFC and influence the 387 system’s ability to achieve treatment goals. The electron 388 equivalents recovered from the cathode during 389 denitrification averaged 94.9 ± 3.9%. 390 391 Inhibition of Nitrite Reduction by Environmental Factors 392

Inhibition of nitrite reduction can be caused by several 393 factors such as insufficient electron donor, insufficient 394 carbon source, presence of dissolved oxygen and pH.24,25,40 395 To determine if insufficient electron donor was a limiting 396 factor causing accumulation of nitrite, the acetate feed in 397 the anode was increased from 154 mg-COD/L.day to 770 mg-398 COD/L.day (Table 2, E4). Despite the increase in the electron 399 donor loading rate, peak nitrite accumulation of 66.2 % of 400 the total nitrogen added was observed. 401

Oxygen inhibition was also considered. Unanticipated 402 oxygen diffusion occurring through the fittings or tubing 403 could have compromised anoxic conditions. In addition to 404 the initial N2 purge, the feed was continuously sparged with 405 nitrogen gas over the course of a batch cycle to maintain 406 anoxic conditions (Table 2, E5). A peak nitrite accumulation 407 of 62.7 % of the total nitrogen added was observed, similar 408 to cycles without a nitrogen-purge. Insufficient carbon 409 source was considered as a possible cause of nitrite 410 accumulation, so the cathode media was supplemented with 411 bicarbonate (Table 2, E6). A peak nitrite accumulation of 412 77.5 % of the total nitrogen was measured. Cathode media 413 with pHs of 6.6, 7.0, and 7.4 were fed to the cathode and the 414 nitrite reduction rate was monitored (Table 2, E7-E9). No 415 significant changes were observed. Changing the 416 environmental conditions to overcome potential 417

denitrification inhibition did not change nitrite accumulation 418 in batch cycles of the cathode. 419

420 Modeling Denitrification 421

Modeling denitrification can improve our understanding of 422 the microbial processes and yield kinetic parameters, which 423 can be useful in designing denitrifying biocathode MFCs. 424 When it was observed that various environmental factors 425 did not significantly affect the accumulation of nitrite in the 426 cathode, it was hypothesized that the accumulation of 427 nitrite was caused due to intracellular electron competition 428 between the enzymes involved in the different steps of 429 denitrification. The ASM-ICE model has been used previously 430 to simulate accumulation of nitrite in suspended cultures 431 and bioreactors. For a denitrifying biocathode, we 432 integrated the Nernst-Monod model, to simulate electron 433 transfer from the electrode to the denitrifying biofilm, into 434 the ASM-ICE (ASM-NICE). ASM-NICE (Table 1) was used to 435 simulate the accumulation of nitrite in the biocathode using 436 a genetic algorithm for estimation of parameters. A Nernst-437 Monod current-potential dependency was observed with a 438 mid-point potential (EkC) of -0.13 V. A mid-point potential of 439 -0.18 V has been previously reported for a denitrifying 440 biocathode.31 It has been previously shown, for bioanodes, 441 that a number of factors, including the source of the 442 inoculum, can influence the value of the mid-point 443 potential.41 444

The kinetic parameters were estimated by calibrating the 445 model to two datasets: the typical batch data (Table 1, E2) 446 and nitrite-only data (Table 1, E3). All the estimated 447 parameter values (Table 3) are in the range of reported 448 values in literature. 27,28 It should be noted that since this is 449 the first study to report kinetic rate constants using ASM-450 NICE, a direct comparison could not be made. KMredNA and 451 KMredNI are affinity constants of the nitrate and nitrite 452 reductase enzymes for the reduced mediator, Mred. The 453 lower the value of these constants, the higher the affinity of 454 the enzyme is to the reduced carrier. The value for KMredNI for 455 nitrite reduction is lower than that for KMredNA for nitrate 456 reduction, indicating that nitrite reduction has a higher 457 capability to compete for electrons when the electron 458


Env

iron

men

talS

cien

ce:W

ater

Res

earc

h&

Tech

nolo

gyA

ccep

ted

Man

uscr

ipt

Publ

ishe

d on

11

Janu

ary

2016

. Dow

nloa

ded

on 1

2/01

/201

6 12

:57:

13.




PAPER




Table 3: Kinetic Parameters for Denitrification in a MFC Biocathode 459

Parameter Source Value Parameter Source Value

𝑟𝑁𝐴_𝑚𝑎𝑥

(mg-N/gVSS•h)

a 1.68 𝑗𝑚𝑎𝑥

(A/m2)

b -0.31

𝐾𝑁𝐼

(mg-N/L)

a 0.56 𝐸𝐾𝑐

(V)

b -0.13

𝑟𝑁𝐼_𝑚𝑎𝑥

(mg-N/gVSS•h)

a 0.45 𝐶𝑡𝑜𝑡

(mmol/gVSS)

c 0.01

𝐾𝑚𝑟𝑒𝑑𝑁𝐴

(mmol/gVSS)

a 0.012 𝐾𝑀𝑜𝑥

(mmol/gVSS)

c 0.0001

𝐾𝑚𝑟𝑒𝑑𝑁𝐼

(mmol/gVSS)

a 0.002 𝐾𝑁𝐴

(mg-N/L)

d 0.0448

Xf

(g-VSS)

b 0.641

a- Parameters estimated in this study b- Experimentally measured parameters 460

c- Assumed values based on Pan et al. 2013 d- Assumed value from Claus and Kutzner 33461

462

Figure 3: Measured (points) and Predicted (lines) Concentrations of Nitrate and Nitrite using ASM-NICE for the calibration Dataset- A) E2 B) E3463

464

465


Env

iron

men

talS

cien

ce:W

ater

Res

earc

h&

Tech

nolo

gyA

ccep

ted

Man

uscr

ipt

Publ

ishe

d on

11

Janu

ary

2016

. Dow

nloa

ded

on 1

2/01

/201

6 12

:57:

13.







466 Figure 4: Measured (points) and Predicted (lines) Concentrations of Nitrate and Nitrite using ASM-NICE for the validation dataset (A) Reactor 1 (B) Reactor 2 467

468 donor is limiting. Since current from the anode remained 469 relatively constant and the concentration of nitrate was high 470 during the initial stages of the batch, nitrate reduction 471 consistently preceded nitrite reduction. Model fits from 472 calibration (Figure 3 and Figure S1) for two different 473 experimental scenarios (E2 and E3) show good agreement 474 with observed data. Model validation was performed with 475 duplicate batch data (E1) from duplicate reactors. The 476 results of the model validation (Figure 4) suggest that the 477 ASM-NICE model for the biocathode is able to simulate the 478 autotrophic denitrification process using the cathode as the 479 electron donor reasonably well (Figure S2). By validating the 480 model with data from two separate MFCs, the validity of the 481 model and the estimated kinetic parameters are 482 demonstrated for different reactors acclimated under 483 similar conditions. Denitrification using a biocathode has 484 been widely used since nitrate has a relative metabolic 485 potential close to that of oxygen (E°NO3-=0.74 V vs SHE, 486 E°O2=0.9 V vs SHE). Using nitrate instead of oxygen 487 eliminates oxygen diffusion across to the anode and thus a 488 source of loss in coulombic efficiencies in MFCs.9 489

In this study, we calibrated and modeled denitrification 490 in a biocathode using ASM-NICE. Such an integrated model 491 has not been presented before to the best of our knowledge. 492 The modeling suggested that when nitrate is at higher 493 concentrations than nitrite, the reduction of nitrite is 494 retarded by the competition for intracellular electron 495 mediators. This would suggest that, when designing 496 continuous-flow biocathodes for nitrate-nitrogen removal, 497 longer hydraulic retention times (HRTs) could resolve the 498 nitrite accumulation. A preliminary sensitivity analysis of the 499

model to the kinetic parameters (KNI, KMox, KNA, EkC, jmax) 500 revealed that these parameters did not have a significant 501 effect on the accumulation of nitrite (data not shown). 502 However, the model showed significant sensitivity to KMredNI 503 and KMredNA (Figure S4 and S5). Briefly, an increase in KMredNI 504 caused the accumulation of nitrite to increase and vice 505 versa. However, reduction of nitrate was not affected. A 506 change in KMredNA affected both nitrate and nitrite reduction 507 since nitrite is formed from the reduction of nitrite. KMredNA 508 and KMredNI are properties of nitrate and nitrite reductase 509 enzymes respectively. The developed model is also able to 510 simulate nitrate and nitrite concentration profiles in a 511 denitrifying biocathode, yielding kinetic parameters (KNI, 512 rNA_max, rNI_max, KMredNA, KMredNI) that can be used for process 513 design. 514

Conclusions 515

This study focused on the dynamics of denitrification in a 516 MFC biocathode, with respect to the accumulation of nitrite 517 before complete denitrification was observed. It was also 518 observed that the power production during nitrate-nitrite 519 reduction was higher compared to that during nitrite 520 reduction. Improvement or control of environmental 521 parameters that affect denitrification pathways did not 522 affect the amount of nitrite accumulation. Denitrification in 523 the biocathode was modeled using ASM-NICE to simulate 524 the use of the cathode electrode as the electron donor. 525 Calibration of the model yielded kinetic parameters (KNI, 526 rNA_max, rNI_max, KMredNA, KMredNI), which could be used for 527 prediction of the performance of a biocathode. The model 528 will serve as a platform for future research into biocathodes 529


Env

iron

men

talS

cien

ce:W

ater

Res

earc

h&

Tech

nolo

gyA

ccep

ted

Man

uscr

ipt

Publ

ishe

d on

11

Janu

ary

2016

. Dow

nloa

ded

on 1

2/01

/201

6 12

:57:

13.







and optimization of their performance. The use of this new 530 model to simulate denitrifying biocathodes under various 531 experimental conditions could yield important information 532 for translating lab-scale studies to pilot scale and full-scale 533 treatment systems. Furthermore, experimental work is 534 needed to determine the specific microorganisms 535 performing autotrophic denitrification in a biocathode and 536 the influence of EKc on their performance. 537

Acknowledgements 538

The authors would like to thank Elizabeth Isenstein for her 539 help with the genetic algorithm. We would also like to thank 540 the Edwin Sisson Fellowship, which funded Varun Srinivasan 541 during the course of this work and start-up funding from the 542 University of Massachusetts-Amherst. 543 544

References 545

546

1 P. Clauwaert, P. Aelterman, T. H. Pham, L. De 547 Schamphelaire, M. Carballa, K. Rabaey and W. 548 Verstraete, Appl. Microbiol. Biotechnol., 2008, 79, 901–549 913. 550

551

2 A. E. Franks and K. P. Nevin, Energies, 2010, 3, 899–919. 552 553

3 B. E. Logan, B. Hamelers, R. Rozendal, U. Schröder, J. 554 Keller, S. Freguia, P. Aelterman, W. Verstraete and K. 555 Rabaey, Environ. Sci. Technol., 2006, 40, 5181–5192. 556

557

4 B. E. Logan, Appl. Microbiol. Biotechnol., 2010, 85, 558 1665–71. 559

560

5 K. B. Gregory and D. R. Lovley, Environ. Sci. Technol., 561 2005, 39, 8943–7. 562

563

6 K. B. Gregory, D. R. Bond and D. R. Lovley, Environ. 564 Microbiol., 2004, 6, 596–604. 565

566

7 S. M. Strycharz, T. L. Woodard, J. P. Johnson, K. P. 567 Nevin, R. A. Sanford, F. R. Loffler and D. R. Lovley, Appl. 568 Environ. Microbiol., 2008, 74, 5943–5947. 569

570

8 C. J. Castro, J. E. Goodwill, B. Rogers, M. Henderson and 571 C. S. Butler, J. Water, Sanit. Hygeine Dev., 2014, 4, 663–572 671. 573

574

9 C. S. Butler and R. Nerenberg, Appl. Microbiol. 575 Biotechnol., 2010, 86, 1399–1408. 576

577

10 P. Clauwaert, K. Rabaey, P. Aelterman, L. de 578 Schamphelaire, T. H. Pham, P. Boeckx, N. Boon and W. 579 Verstraete, Environ. Sci. Technol., 2007, 41, 3354–60. 580

581

11 B. Virdis, K. Rabaey, R. a Rozendal, Z. Yuan and J. Keller, 582 Water Res., 2010, 44, 2970–80. 583

584

12 N. Guerrero-Rangel, J. A. Rodriguez-de la Garza, Y. 585 Garza-Garcia, L. J. Rios-Gonzalez, G. J. Sosa-Santillan, I. 586 M. de la Garza-Rodrguez, S. Y. Martinez-Amador, M. M. 587 Rodriguez-Garza and J. Rodriguez-Martinez, Int. J. 588 Electr. Power Eng., 2010, 4, 27–31. 589

590

13 S. Pandit, A. Sengupta, S. Kale and D. Das, Bioresour. 591 Technol., 2011, 102, 2736–2744. 592

593

14 A. M. Fan and V. E. Steinberg, Regul. Toxicol. 594 Pharmacol., 1996, 23, 35–43. 595

596

15 G. Ciudad, O. Rubilar, P. Muñoz, G. Ruiz, R. Chamy, C. 597 Vergara and D. Jeison, Process Biochem., 2005, 40, 598 1715–1719. 599

600

16 B. E. Rittmann and P. L. McCarty, Environmental 601 Biotechnology: Principles and Applications, New 602 York:McGraw-Hill, 2001. 603

604

17 P. Clauwaert, J. Desloover, C. Shea, R. Nerenberg, N. 605 Boon and W. Verstraete, Biotechnol. Lett., 2009, 31, 606 1537–1543. 607

608

18 J. Desloover, S. Puig, B. Virdis, P. Clauwaert, P. Boeckx, 609 W. Verstraete and N. Boon, Environ. Sci. Technol., 2011, 610 45, 10557–66. 611

612

19 S. Puig, M. Serra, A. Vilar-Sanz, M. Cabré, L. Bañeras, J. 613 Colprim and M. D. Balaguer, Bioresour. Technol., 2011, 614 102, 4462–7. 615

616

20 T. Van Doan, T. K. Lee, S. K. Shukla, J. M. Tiedje and J. 617 Park, Water Res., 2013, 47, 7087–97. 618

619

21 W. M. J. Lewis and D. P. Morris, Trans. Am. Fish. Soc., 620 1986, 115, 183–195. 621

622

22 C. S. Bruning-Fann and J. B. Kaneene, Vet. Hum. 623 Toxicol., 1993, 35, 521–38. 624

625

23 IPCC, J. T. Houghton, Y. Ding, D. J. Griggs, M. Noguer, P. 626 J. van der Linden, X. Dai, K. Maskell and C. A. Johnson, 627 Climate Change 2001: The Scientific Basis, Cambridge 628


Env

iron

men

talS

cien

ce:W

ater

Res

earc

h&

Tech

nolo

gyA

ccep

ted

Man

uscr

ipt

Publ

ishe

d on

11

Janu

ary

2016

. Dow

nloa

ded

on 1

2/01

/201

6 12

:57:

13.







University Press, Cambridge, United Kingdom and New 629 York, NY, USA, 2001. 630

631

24 H. Körner and W. G. Zumft, Appl. Environ. Microbiol., 632 1989, 55, 1670–1676. 633

634

25 Y. Zhou, A. Oehmen, M. Lim, V. Vadivelu and W. J. Ng, 635 Water Res., 2011, 45, 4672–82. 636

637

26 W. C. Hiatt and C. P. L. Grady, Water Environ. Res., 638 2008, 80, 2145–2156. 639

640

27 Y. Pan, B. J. Ni and Z. Yuan, Environ. Sci. Technol., 2013, 641 47, 11083–11091. 642

643

28 Y. Pan, B.-J. Ni, H. Lu, K. Chandran, D. Richardson and Z. 644 Yuan, Water Res., 2015, 71, 21–31. 645

646

29 A. K. Marcus, C. I. Torres and B. E. Rittmann, Biotechnol. 647 Bioeng., 2007, 98, 1171–1182. 648

649

30 A. Ter Heijne, O. Schaetzle, S. Gimenez, F. Fabregat-650 Santiago, J. Bisquert, D. P. B. T. B. Strik, F. Barrière, C. J. 651 N. Buisman and H. V. M. Hamelers, Energy Environ. Sci., 652 2011, 4, 5035. 653

654

31 K. P. Gregoire, S. M. Glaven, J. Hervey, B. Lin and L. M. 655 Tender, J. Electrochem. Soc., 2014, 161, H3049–H3057. 656

657

32 C. I. Torres, A. K. Marcus, P. Parameswaran and B. E. 658

Rittmann, Environ. Sci. Technol., 2008, 42, 6593–6597. 659 660

33 G. Claus and H. J. Kutzner, Appl. Microbiol. Biotechnol., 661 1985, 22. 662

663

34 K. Soetaert, T. Perzoldt and W. R. Setzer, J. Stat. Softw., 664 2010, 33, 1–25. 665

666

35 R Core Team, R: A language and environment for 667 statistical computing., R Foundation for Statistical 668 Computing, Vienna, Austria, 2015. 669

670

36 H. Wickham, ggplot2: Elegant graphics for data 671 analysis, Springer, New York, 2009. 672

673

37 G. J. Pelletier, S. C. Chapra and H. Tao, Environ. Model. 674 Softw., 2006, 21, 419–425. 675

676

38 K. Deb, A. Pratap, S. Agarwal and T. Meyarivan, IEEE 677 Trans. Evol. Comput., 2002, 6, 182–197. 678

679

39 B. Virdis, K. Rabaey, Z. Yuan and J. Keller, Water Res., 680 2008, 42, 3013–24. 681

682

40 R. Knowles, Microbiol. Rev., 1982, 46, 43–70. 683 684

41 J. F. Miceli, P. Parameswaran, D.-W. Kang, R. 685 Krajmalnik-Brown and C. I. Torres, Environ. Sci. 686 Technol., 2012, 46, 10349–55. 687

688 689


Env

iron

men

talS

cien

ce:W

ater

Res

earc

h&

Tech

nolo

gyA

ccep

ted

Man

uscr

ipt

Publ

ishe

d on

11

Janu

ary

2016

. Dow

nloa

ded

on 1

2/01

/201

6 12

:57:

13.



view article online envonir ment l a...

Documents