See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/316476744

Enhancing denitrification using a novel in situ membrane biofilm reactor

(isMBfR)

Article  in  Water Research · April 2017

DOI: 10.1016/j.watres.2017.04.054

CITATIONS READS

4 129

7 authors, including:

Yonghong Wu Aura Ontiveros-Valencia

Institute of Soil Science, Chinese Academy of Sciences, China Tecnológico de Monterrey
111 PUBLICATIONS   992 CITATIONS    28 PUBLICATIONS   313 CITATIONS   

SEE PROFILE SEE PROFILE

Luis ALBERTO Ordaz-Diaz Junzhuo Liu

Universidad Politécnica de Durango Carretera Durango Chinese Academy of Sciences
11 PUBLICATIONS   16 CITATIONS    24 PUBLICATIONS   170 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Bioenergy View project

PhycoPigments: Novel manufacturing methods for high value pigments products from microalgae View project

All content following this page was uploaded by Yonghong Wu on 13 March 2018.

The user has requested enhancement of the downloaded file.

 Water Research 119 (2017) 234e241

Contents lists available at ScienceDirect

Water Research
journal homepage: www.elsevier.com/locate/watres

Enhancing denitrification using a novel in situ membrane biofilm

reactor (isMBfR)
Yonghong Wu a, b, *, Yizhou Li b, Aura Ontiveros-Valencia b, Luis Ordaz-Díaz b,
Junzhuo Liu a, Chen Zhou b, Bruce E. Rittmann b
a
State Key Laboratory of Soil and Sustainable Agriculture, Institute of Soil Science, Chinese Academy of Sciences, 71 East Beijing Road, Nanjing 210008, PR
China
b
Biodesign Swette Center for Environmental Biotechnology, Arizona State University, P. O. Box 875701, Tempe, AZ 85287-5701, USA

a r t i c l e i n f o a b s t r a c t

Article history: The insufficient supply of electron donor in surface water contaminated with nitrate leads to its
Received 4 January 2017 incomplete reduction in natural or constructed wetlands. Although the addition of organic matter
Received in revised form (represented as chemical oxygen demand, COD) can stimulate N removal by denitrification, direct
18 April 2017
supplementation of COD creates unacceptable risks to effluent quality. An alternative for stimulating
Accepted 23 April 2017
Available online 25 April 2017
denitrification is supplying hydrogen gas (H2) as an inorganic electron donor. We evaluate an innovative
means to do H2-based denitrification of surface waters in a wetland setting: the in-situ membrane
biofilm reactor (isMBfR), in which H2 is delivered to a biofilm of denitrifying bacteria on demand based
Keywords:
Membrane biofilm reactor
on the presence of nitrate. We carried out a proof-of-concept study in which an upper “photo zone” and a
Chemical oxygen demand lower “MBfR root zone” were combined to remove nitrate and COD from simulated surface water.
Nitrate reduction Employing mass-balances for H2, COD, nitrate, and oxygen, we documented nearly complete removals of
Hydrogen nitrate and COD, except when the H2 supply was intentionally shut off. All nitrate removal was
accomplished in the “MBfR root zone,” where H2 delivery supplemented the COD supply (as needed) and
provided the large majority of electron equivalents to reduce nitrate to N2.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction situation in a wetland is that N accumulates in the form of NO3 , but

Wetlands contribute to surface-water quality and are natural
habitats for fauna and flora (Drayer and Richter, 2016; Yang et al.,
2016). Whether the wetland is natural or constructed, it can
remove organic carbon (OC) and nitrogen (N) (Al-Isawi et al., 2017;
Wu et al., 2010). The degrees to which OC and N coming to the
wetlands can be removed depend on how the supply of electron
donors (chemical oxygen demand (COD) and NHþ 4 ) is balanced by
the input of electron acceptors (O2 and NO 3 ) (Ifabiyi, 2008;
Vagnetti et al., 2003). When a wetland is used for wastewater
treatment, the influent generally contains considerable COD and
NHþ 
4 , but relatively less O2 and NO3 (Allen et al., 2010; Badhe et al.,
2014). However, O2 enters the water by surface aeration and
photosynthesis (Allen et al., 2010; Badhe et al., 2014). With dis-
solved O2 (DO) available, NHþ 4 can be nitrified to form NO2 , and


NO3 , and COD is aerobically oxidized. As a result, the common
* Corresponding author. 71, East Beijing Road, Nanjing 210008 China.
E-mail addresses: yhwu@issas.ac.cn (Y. Wu), Rittmann@asu.edu (B.E. Rittmann).
little COD remains to drive microbial denitrification.
For this typical scenario, complete N removal requires that an
electron donor be supplemented. While supplementing an organic
electron donor is an option (Moussavi et al., 2015; Scherson et al.,
2013), its addition to a wetland ecosystem may cause negative
impacts, such as excessive formation of microbial biomass,
increased turbidity, sediment build up, severe DO depletion, and
residual COD in the effluent (Rittmann et al., 2005).
Hydrogen gas (H2) is inorganic electron donor, and it has been
extensively studied for end-of-pipe denitrification using the
membrane biofilm reactor (MBfR) (Hasar et al., 2008; Lee and
Rittmann, 2002; Tang et al., 2011; Xia et al., 2010). Compared to
adding an organic donor, using H2 offers numerous advantages:
relatively low cost per electron delivered; not toxic to humans; low
productions of biomass, turbidity, and sediment; no donor residual;
less expensive than most organic donors; and can be delivered “on
demand” with essentially 100% efficiency (Lee and Rittmann, 2002;
Rittmann et al. 2004, 2005; Xia et al., 2011; Zhou et al., 2016).
For end-of-pipe treatment, the H2-based MBfR has been applied

http://dx.doi.org/10.1016/j.watres.2017.04.054
0043-1354/© 2017 Elsevier Ltd. All rights reserved.
 Y. Wu et al. / Water Research 119 (2017) 234e241 235

to reduce one or several oxidized contaminants, including nitrate,
perchlorate, selenate, chromate, vanadium (V), and uranium from
waters having minimal or no COD (Chung et al., 2006; Ontiveros-
Valencia et al., 2012; Rittmann, 2006, 2007; Rittmann et al.,
2005; Xu et al., 2015; Zhao et al., 2011; Zhou et al., 2014, 2016). A
biofilm dominated by H2-oxidizing autotrophs accumulates on the
exterior surface of non-porous hollow-fiber membranes. H2 dif-
fuses through the membrane wall and is consumed within the
biofilm; the H2 concentration in the bulk water is negligible (Lee
and Rittmann, 2002; Ziv-El and Rittmann, 2009).
In a wetland setting, particularly when used for wastewater
treatment, organic matter (i.e., COD) co-exists with NHþ 
4 and NO3
(Wu et al., 2011a, 2011b; Zhao et al., 2016). Thus, the H2-based MBfR
has to function well when COD oxidation must be achieved in
parallel. Hybrid aerobic/anoxic MBfR systems based on H2 and O2
delivery in integrated MBfR units successfully removed total N
(Cowman et al., 2005; Hasar et al., 2008). However, in the wetland
setting, O2 delivery takes place by the usual means of advection in
the influent, surface aeration, and photosynthesis. Thus, the role of
the MBfR is to enhance denitrification that occurs in parallel with
aerobic COD oxidation and nitrification.
The main objective of this study was to demonstrate that a H2-
based in situ MBfR (isMBfR) could augment denitrification in par-
allel with COD oxidation in synthetic wastewater with low COD/TN
ratio. To achieve this goal, we examined the removal efficiencies of
TN and COD in a bench-scale simulation of how an isMBfR can be
integrated into a wetland.
2. Materials and methods
2.1. The isMBfR design
To test the isMBfR in a simple wetland scenario, a bench-scale
system was divided into a top “photo zone” and a bottom “MBfR
root zone.” As illustrated in Fig. 1, a phototrophic top zone was
simulated by using 150 cm3 of low-density (18 kg/m3) sponges; it
had a volume of 210 cm3 and was illuminated. The MBfR root zone
held 7 bundles (with a total surface area of 0.07 m2) of poly-
propylene fibers (products of Teijin Fibers Ltd., Osaka, Japan) sealed
into H2 ports to deliver H2. The other ends of the fibers were shut
with glue. Each bundle consisted of 12 fibers of 14-cm length.
During experimentation, H2 was supplied from the top through
tubing and distributed to the 7 H2 ports. The H2 was supplied using
a H2 cylinder, and the H2 pressure to the fibers using a pressure
regulator. The volume of the bottom compartment (below the
floating support and holding the isMBfR) was ~450 mL. Separating
the top and bottom zones was a floating spacer of 2-cm thick
polyethylene. A stir bar (325 rpm) mixed the contents of the bottom
section, but mixing with the top photo zone was negligible. Tubing
connected the outlet of the “photo-zone” removed effluent from
the system.
2.2. Start-up and operating conditions
Initially, the isMBfR was inoculated with 30 ml of activated
sludge from the Mesa Northwest Wastewater Reclamation Plant
(Mesa, AZ, USA). The inoculation liquid was held in the system for 4
days to establish a biofilm, at which time the inoculum was flushed
from the system before continuous feeding began.
The isMBfR was fed with synthetic wastewater with the
following inorganic composition (per liter) for the standard con-
dition: 0.087 g NaNO3, 0.252 g NaHCO3, 0.0053 g KH2PO4, 0.050 g
MgSO4$7H2O, and 1 ml trace mineral solution. The trace mineral
solution contained (per liter): 100 mg ZnSO4$7H2O, 30 mg
MnCl2$4H2O, 300 mg H3BO3, 200 mg CoCl2$6H2O, 10 mg
CuCl2$2H2O, 10 mg NiCl2$6H2O, and 30 mg Na2SeO3. The total

Fig. 1. Schematic of the isMBfR (height х diameter ¼ 23.0 х 4.0 cm) used in this study. The flow was once-through. The influent (including nutrients and contaminants (i.e. organic
materials and nitrate)) was delivered from a single medium tank during stages 1e4 (without COD supply). To minimize denitrification in the influent medium, the nutrients and the
contaminants were input individually from two medium tanks (not shown here) from stage 5e7 (with COD supply). The light source was a table lamp delivering 875 lumens light
intensity, and it stimulated the accumulation of photosynthetic microorganisms on the sponges in the photo zone.
236 Y. Wu et al. / Water Research 119 (2017) 234e241

Table 1
The operating and influent conditions in the 7 stages.

Stage 1 Stage 2 Stage 3 Stage 4 Stage 5 Stage 6 Stage 7

H2 pressure (psiga) 14 12 15.5 15 15 15 0

Flow rate (L/d) 0.28 0.26 0.26 0.30 0.30 0.30 0.30
EBHRT (d) 2.1 2.3 2.3 2.0 2.0 2.0 2.0
Influent pHb e 8.4 6.97 ± 0.07 6.99 ± 0.09 7.03 ± 0.06 7.13 ± 0.05 7.29 ± 0.03
Nitrate concentration in inlet (mg-N/L)b 15.0 ± 0.3 13.5 ± 0.3 12.4 ± 0.6 14.5 ± 0.7 13.7 ± 0.4 13.3 ± 0.4 13.7 ± 0.1
COD concentration in the inlet (mg/L)b 0 0 0 0 10.2 ± 1.8 14.7 ± 1.1 11.9 ± 0.4
DO concentration in the inlet (mg/L) 2.9 1.7 1.5 1.2 2.7 3.4 4.1
a
For stages 1e6, the absolute H2 pressure in the fiber lumen was (14.7 þ psig)/14.7 atm. For stage 7, the H2 absolute pressure was zero.
b
The values refer to means ± standard deviation.

nitrogen (TN) was NO 3 at a concentration of 14 mg-N/L; thus, the

experiments simulated a situation in which nitrification was ach-
ieved in an upstream treatment process or wetland. Organic com-
ponents were an equal-COD mixture of sodium acetate
(CH3COONa), sodium lactate (C3H5O3Na), and citrate (C7H6O2). The
mixed influent was sterilized by the autoclaving for 30 min, and
then the pH was adjusted to 7.0 ± 0.5 using 1 N H2SO4.
The influent entered into the bottom MBfR root zone and then
moved to the top photo-zone. Effluent exited from the photo zone.
The total empty-bed hydraulic retention time (EBHRT) in the
isMBfR (the entire reactor) was approximately 2 d. Operation of the
isMBfR was conducted in 7 stages that had different influent con-
centrations of NO3 and COD, as well as H2 pressure to the mem-
Fig. 2. The mass balances for COD and nitrate in the isMBfR. Q, flow rate (L/d); Cin,
brane fibers. Table 1 summarizes the operating and influent concentration of COD in the influent (g COD/L); Ce, concentration of COD in the effluent
conditions for the seven stages. The influent in Stages 1e4 had no (g COD/L); Sin, concentration of nitrate in the influent (g N/L); Se, concentration of
COD, which means that the isMBfR had to be responsible for all nitrate in the effluent (g N/L); b, O2 consumed for COD respiration (g O2/d); and c, H2
denitrification. In Stages 5e7, the influent contained COD, which consumed nitrate respiration (g N/d).

means that the isMBfR had to supplement denitrification driven by

COD oxidation.
2.3. Samples and analyses
Water samples, collected daily, were immediately filtered
through a 0.2-mm membrane filter (Pall Corp., Ann Arbor, ML). Ion
Chromatography (ICS-3000 RFIC, Dionex Corp.) was used for
measuring the ionic species: an AS18 Anion-Exchange Column for
anions nitrate and nitrite and a CS12A Cation-Exchange Column for
ammonium. Total nitrogen (TN) was the sum of NO 
3 -N, NO2 -N, and
NHþ4 -N. COD was analyzed by HACH COD kits (Colorado, USA) with
ranges of 20e1500 and 0e60 mg/L. The pH and dissolved oxygen
(DO) in water were determined using pH and DO probes (Thermo
Electron Corporation).
2.4. Mass balances at steady state
Fig. 2 summarizes the mass balances for COD and nitrate, and it
defines the terms. In brief, COD could be oxidized through respi-
ration of O2 and nitrate, while nitrate could be the electron acceptor
for COD and H2 oxidations. Thus, the mass per time of nitrate-
consumed by COD oxidation was QCin -b -QCe (g COD/d), where b
is the mass rate of COD consumed by O2 respiration. Similarly, the
mass per time of COD consumed by nitrate respiration was QSin -c
-QSe (g COD/d), where c is the mass rate of H2 consumed for nitrate
respiration. We were able to measure Q, Cin, Ce, Sin, and Se, which
allowed us to compute b and c.
The mass/time rates for nitrate and COD removals (R, in mg-N/
d for nitrate and mg-COD/d for COD) for the “MBfR root zone”
and “photo zone” were computed with Eq. (1).
R ¼ Q  DCOD ðor Dnitrate Þ
Q is the volumetric flow rate (L/d), and DCOD ðor Dnitrate Þ is the
difference between the average influent and effluent COD (or ni-
trate) concentrations of each stage for each compartment (mg-
COD/L for COD, mg-N/L for nitrate).
To model the relationships among H2, COD, and nitrate, three
assumptions were made: 1) The H2 supplied to the system and the
organics in the influent are entirely utilized for oxygen respiration
and denitrification; this was true because no other electron ac-
ceptors were present. 2) The amount of oxygen produced by
photosynthesis or transferred from atmosphere was negligible in
the “MBfR root zone,” because mixing was negligible between the
two zones. 3) Denitrification based on oxidation of any electron
donors generated by phototrophs in the “photo zone” was negli-
gible, again because mixing between zones was negligible.
We first built the stoichiometric relationship between nitrate
and H2 with no COD input. According to Tang et al. (2012), the
maximum delivery flux of H2 (Jmax , g/m2-d) can be estimated based
on the H2 pressure (P, psig) using Eqs. S1 and S2 in Supplementary
Information (SI). For the first 4 experimental stages, H2 was
oxidized by respirations of NO3 and O2; then, the maximum H2 flux
used for denitrification was the maximum total H2 flux (Jmax ) minus
the H2 flux utilized for O2 respiration. The O2 flux was calculated by
Eq. S3 in SI. The relationship between H2 and O2 fluxes was based
on the stoichiometry in Eq. S4, which allowed us to convert the O2
removal flux to the H2 flux for O2 respiration (JO2 H2 , in g-H2/m2-d)
using Eq. S5. Then, the H2 flux for nitrate removal (JNO3 H2 , in g-H2/
m2-d) was calculated from:
Q DDO 1g
JNO3 H2 ¼ Jmax  JO2 H2 ¼ 0:0145P  0:125  
AM 1000mg
(2)
(1)
where DDO is the difference between influent and bulk DO con-
centration (mg/L). AM is the membrane surface area (m2).
 Y. Wu et al. / Water Research 119 (2017) 234e241 237

To build the stoichiometric relationship between consumption intermediate fraction for biomass net synthesis during the het-
rates for nitrate and H2, we assumed that the percentage of electron erotrophic denitrification. The stoichiometries for nitrate respira-
donor for cell synthesis (fs) ranged from 0 to 0.25 (Rittmann and tion with the 3 organic materials (i.e., acetate, lactate, and citrate)
McCarty, 2001). fs ¼ 0 means no biomass formation in the deni- utilized in this experiment are shown in Eqs. S(6) e S8. Based on
trification process, while fs ¼ 0.25 indicates a maximum percentage Eqs. S(8) e S10, the COD removed by nitrate respiration (RCODN , in
of biomass formation in an autotrophic denitrification process mg-N/d) was calculated using Eq. S9.
(Rittmann and McCarty, 2001). When fs ¼ 0, all H2 is used for nitrate The total mass/time nitrate removal (RT;N , in mg-N/d) was
reduction, and this gives the maximum nitrate-removal flux for a calculated by summing the COD and H2 consumptions for nitrate
certain H2 supply; when fs ¼ 0.25, the minimum percentage of H2 is respiration (RH2 N , in mg-N/d) from Eq. (7):
used for nitrate reduction, because some of the H2 is used for
reducing CO2 and NO 3 for biomass synthesis. The actual relation- RT;N ¼ RCODN þ RH2 N
ship had to lie between these two limits. Q  ðDCOD  DDO Þ Q Dnitrate 1g
We chose fs ¼ 0.1 in this study, because MBfR biofilms typically ¼ þ  (7)
3:12 AM 1000mg
have a large average solids retention time and significant endoge-
nous decay (Rittmann, 2007; Rittmann and McCarty, 2001). Then, where DCOD is the difference between the influent and bulk COD
the stoichiometry is described by Eq. (3): concentration of “MBfR root zone” (mg/L).
Table 2 summarizes the parameters needed for the calculations
0:159NO þ
3 þ 0:159H þ 0:5H2 þ 0:045CO2 for stages 1e7.
¼ 0:55H2 O þ 0:009C5 H7 O2 N þ 0:075N2 (3)
3. Results and discussion
Based on the stoichiometry of Eq. (3), we converted the
maximum H2 flux (g/m2-d) for nitrate respiration to the maximum
3.1. Nitrate and COD removals in the isMBfR
estimated nitrate removal flux (JNO3 est g-N/m2-d) using Eq. (4):

JNO3 H2 In the first 4 stages, no organic matter was supplied to the
JNO3 est ¼ (4) system, making H2 the only exogenous electron donor. Fig. 3
0:45
summarizes the nitrate concentrations in the influent and
where 0.45 is the stoichiometric conversion from g H2 to g NO3-N effluent of the isMBfR during the first 4 stages. From the beginning
from Eq. (3). of Stage 1, the effluent NO 3 concentration was below 4 mg-N/L.
Substituting Eq. (4) into Eq (2) gives the JNO3-est in terms of P and Nearly 100% nitrate removal was achieved at the end of Stage 1.
DDO: The H2 pressure was adjusted from 14 psig to 12 psig
(1.95e1.82 atm absolute pressure) at the beginning of Stage 2, and
 
this led to 93% N removal. At the end of Stage 2, a biofilm sample
0:0145P  0:125  QADMDO  1000mg
1g
JNO3 est ¼ (5) was collected by cutting off a piece of fiber. This disturbance caused
0:45 the nitrate concentration to increase (to ~ 2.7 mg-N/L), and the H2
It can be compared to the experimental nitrate-removal flux pressure was raised to 15.5 psig (2.05 atm) to help recover nitrate
(JNO3 exp , in g-N/m2-d) calculated with Eq. (6): removal in Stage 3. The effluent concentration of nitrate was be-
tween 1 and 4 mg-N/L in Stage 3, and another biofilm sample was
 
taken at the end of Stage 3.
Q S0nitrate  Snitrate 1g
JNO3 exp ¼  Stage 4 had a relatively long duration (39 days) to ensure that
AM 1000mg the biofilm had enough time to recover from the disturbance
Q Dnitrate 1g caused by the biofilm sampling at the end of Stage 3. In addition,
¼  (6)
AM 1000mg the flow rate was increased by 15% to increase the nitrate loading.
The effluent concentration spiked from the 61st day to the 65th day
S0nitrate is the influent nitrate concentration (mg-N/L), Snitrate is the due to an accidental decrease of the H2 pressure on day 61. After the
bulk concentration for nitrate in “MBfR zone” (mg-N/L), and Dnitrate pressure was corrected (within 2 h), the effluent nitrate concen-
is the difference between influent and bulk concentration of nitrate tration returned to its previous low (~4 mg-N/L, or ~77% removal)
(mg-N/L). by the 68th day.
Stages 5, 6, and 7 had COD supplied in the influent, and deni- The average COD and nitrate concentrations in the influent and
trification could be driven by COD oxidation, as well as by H2 effluent are summarized in Fig. 4 for Stage 5e7, when organic
oxidation. Since heterotrophs tended to be present in the bulk electron donor was supplied in the influent. A relatively low con-
liquid and the outer layer of the biofilm, we assumed that the centration of organic matter (~10.2 mg COD/L) was supplied as an
heterotrophs consumed DO first. We build the relationship be- exogenous electron donor in Stage 5. Then, the influent COD con-
tween nitrate and COD through stoichiometry. We chose fs ¼ 0.1, an centration was increased to ~14.7 mg COD/L in Stage 6. Stage 7 had

Table 2
Parameters needed for the calculations in stages 1e7.

H2 pressure (psig) Max. H2 flux (g-H2/m2-d) Flow rate Q (L/d) Dnitrate (mg-N/L) DDO (mg/L) DCOD (mg/L)
Stage 1 14 0.15 0.28 11.7 2.9 0
Stage 2 12 0.12 0.26 11.9 1.7 0
Stage 3 15.5 0.18 0.26 9.5 1.5 0
Stage 4 15 0.16 0.3 10.0 1.2 0
Stage 5 15 0.16 0.3 10.9 2.7 7.7
Stage 6 15 0.15 0.3 12.0 3.4 12.5
Stage 7 0 0 0.3 0 4.1 10.6
238 Y. Wu et al. / Water Research 119 (2017) 234e241

Fig. 3. Nitrate concentration as a function of time in the first four stages, which had no input organic matter. The H2 pressures for the 4 stages were 14 psig, 12 psig, 15.5 psig, and 15
psig. The flow rates for the 4 stages were 0.28 mL/d, 0.26L/d, 0.26 L/d, and 0.3 L/d.

no H2 supply, and the influent COD concentration was ~11.9 mg state for Stages 2e7; DO was not measured in Stage 1. Due to bio-
COD/L. logical activity in the influent tubing, the influent DO concentration
When H2 and COD were co-supplied as electron donors (Stages was below 4.1 mg/L. The DO concentration was negligible in “MBfR
5 and 6), the effluent nitrate concentration stayed at a low level root zone,” < 0.1 mg/L. Phytoplankton accumulated in the “photo
(<3 mg-N/L, 82e91% N removal). Due to the greater organic elec- zone” produced oxygen through photosynthesis, and the effluent
tron donor in Stage 6 than in Stage 5 (~14.7 mgCOD/L versus ~10.2 DO was around 4 mg/L, but the DO concentration at the bottom of
mgCOD/L), the nitrate concentration in the effluent was lower than the “photo-zone” also was minimal (less than 0.1 mg/L for all but
in Stage 6 (~1.2 mg-N/L, 91% nitrate removal). However, when the
H2 supply was stopped in Stage 7, the effluent nitrate concentration
returned to a high level (~11.5 mg-N/L, only 12% nitrate removal).
The results in Figs. 3 and 4 make it clear that major nitrate
removal was possible when the isMBfR was operating, and this was
true whether or not the influent contained COD. Turning off H2
delivery to the isMBfR resulted in only small removal of nitrate,
because the COD present in the influent was insufficient. Thus, the
isMBfR was able to overcome a major limitation of wetlands for
nitrogen removal.

3.2. Comparison of nitrate removals in the MBfR root-zone and the

photo-zone

Fig. 5 summarizes the average nitrate concentrations at different

locations of the experimental system. “Influent” indicates the
concentrations in the influent to the isMBfR, “MBfR zone” indicates
the bulk concentrations in the “MBfR root zone,” “Photo-zone”
refers to the concentrations in the outlet of the “photo zone,” and
“Effluent” refers to the concentrations at the end of the effluent
tubing. The concentrations were for the averages for the stages.
Compared to the low nitrate-removal of nitrate (~11%) in stage 7
without H2 supply, the high nitrate-removal percentages for the
first 6 stages (72e92%) demonstrate that the isMBfR achieved the
goal of stimulating major NO 3 reduction with or without COD
input. The difference occurred solely in the “MBfR root zone,”
which reinforces that the isMBfR was the core of denitrification for
this bench-scale system.

3.3. Dissolved oxygen in the isMBfR

Fig. 4. Nitrate and COD concentrations, as well as their removal rates, for stages 5e7,
Table 3 summarizes the DO concentrations measured at steady when COD was present in the influent. The flow rate for these 3 stages was 0.3 L/d.
 Y. Wu et al. / Water Research 119 (2017) 234e241 239

Fig. 5. Nitrate concentrations at the different locations.

stage 7, when it was only 1 mg/L). The low DO concentration at the competed for space with the autotrophic denitrifiers, but any
bottom of the “photo zone” means that DO entering the “MBfR root negative impact was small, since NO
3 removal was nearly complete
zone” from the “photo zone” was negligible compared to what in Stage 6.
entered in the influent. Explanations are that the production of In Stage 7, the experimental NO 3 removal flux, solely from
oxygen was small in the “MBfR root zone” and that advection in the heterotrophic oxidation of COD, corresponded well to the
“photo zone” (superficial upflow velocity of 2.3e2.7 cm/d) trans- maximum estimated removal flux based on COD removal. However,
ported DO away from the “MBfR root zone.” the total removal of NO
3 was much less in Stage 7 due to the limited
amount of COD available to drive denitrification.
3.4. Analysis of fluxes
3.5. Practical implications
For the first 4 stages, we compared the maximum nitrate-
The isMBfR is a novel mean to deliver H2 to drive denitrification
removal flux based the H2 pressure (Eq. (5)) with the experi-
in surface water. Aquatic plants and phytoplankton may enhance or
mental nitrate removal flux based on Eq. (6). They are compared in
impair nitrogen removal, but the main role of the isMBfR is to in-
the left set of 4 bar graphs in Fig. 6. For all four stages, the experi-
crease the reduction of NO 3 to N2 beyond what is possible with
mental nitrate-removal flux was lower than the maximum nitrate
only the naturally occurring processes of denitrification and plant
removal fluxes. Since some nitrate was present in the effluent (an
uptake.
average concentration of 2.7 mg-N/L) and sulfate reduction did not
occur (data not shown, as the influent and effluent concentrations
of sulfate were nearly identical), NO 3 reduction was limited by
mass-transport or utilization kinetics of the biofilm.
For Stages 5, 6, and 7, the nitrate-removal flux by H2 and COD
oxidations was based on Eq. (7). The fluxes are compared with the
experimental mass/time nitrate removals in the 3 bar graphs to the
right in Fig. 6. For Stages 5 and 6, the experimental nitrate-removal
fluxes were close to the maximum estimate nitrate removal fluxes,
which indicates that the combined supply of COD and H2 was
sufficient to completely reduce O2 and reduce most NO 3 . Oxidation
of COD means that the autotrophic denitrifiers had to consume less
H2 than in stages with no COD addition. With COD supplied, het-
erotrophic denitrifiers accumulated in the “MBfR root zone,” pre-
sumably on the membrane fibers, since virtually all denitrification
took place in that zone. It is possible that the heterotrophs

Table 3
Steady-state DO concentrations measured at different locations in the isMBfR.

Stage 2 Stage 3 Stage 4 Stage 5 Stage 6 Stage 7

Influent 1.7 1.5 1.2 2.7 3.4 4.1

Fig. 6. Comparison of experimental and estimated nitrate removal fluxed. The esti-
MBfR zone <0.1 <0.1 <0.1 <0.1 <0.1 0.8
mated mass/time nitrate removal from Stage 5e7 was the maximum flux based on the
Bottom of photo-zone <0.1 <0.1 <0.1 <0.1 <0.1 1.0
H2 pressure and influent COD.
240 Y. Wu et al. / Water Research 119 (2017) 234e241
Adhering to the stoichiometry established here between the H2-
delivery capacity and needed nitrate removal is important for
effective utilization of the isMBfR. Most importantly, insufficient
H2-delivery capacity will prevent complete nitrate removal, as
illustrated by Stage 7. However, excess delivery of H2 should be
avoided, since excess delivery of H2 is an unnecessary cost that may
cause sulfate reduction to H2S (Ontiveros et al., 2012; Zhao et al.,
2014). Fortunately, H2 delivery in an on-demand process in the
MBfR, and excess delivery is unlikely unless the fibers are damaged
(Rittmann, 2007; Tang et al., 2012).
In a real wetland, O2 may enter the “MBfR root zone” at a higher
rate than in our bench-scale system. The impacts of O2 addition
may have the greatest impact on the removal of organic matter,
augmenting the COD oxidation and leading to less COD available for
the denitrification. This situation will increase the demand for H2
from the isMBfR, and the stoichiometry relationships developed
here are of practical value for determining the extra H2 demand.
4. Conclusions
The lack of exogenous electron donors usually limits the nitrate-
removal capacity for a constructed or natural wetland. The isMBfR
augments the performance of constructed wetland by overcoming
the inherent deficiency for electron donors by supplying H2 on
demand. This study evaluated the isMBfR for enhanced nitrate
removal from synthetic wastewater and investigated the relation-
ships among nitrate, O2, COD, and H2. The most important finding is
that the isMBfR accomplished nearly complete nitrate reduction
and removal when the COD was insufficient in the influent, the
usual condition in a wetland. For the first 6 experimental stages,
when the isMBfR was active, at least 70% nitrate removal was
achieved, and removals could approach 100%.
Using mass-balance relationships among H2, COD, nitrate, and
DO, we found that the actual nitrate removal corresponded well to
the H2-delivery capacity and COD supply rate in the influent. Most
of the removal was due to H2-delivery with the realistic NO 3 and
COD concentration used here. These results document that aug-
menting the self purification capacity of a wetland with the isMBfR
can lead to nearly complete N removal.
Acknowledgements
This work was also supported by the National Natural Science
Foundation of China (41422111), the State Key Development Pro-
gram for Basic Research of China (2015CB158200) and the Natural
Science Foundation of Jiangsu Province China (BK20150066). This
work was also supported by Youth Innovation Promotion Associa-
tion, Chinese Academy of Sciences (2014269).
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.watres.2017.04.054.
References
Al-Isawi, R., Ray, S., Scholz, M., 2017. Comparative study of domestic wastewater
treatment by mature vertical-flow constructed wetlands and artificial ponds.
Ecol. Eng. 100, 8e18.
Allen, J.G., Beutel, M.W., Call, D.R., Fischer, A.M., 2010. Effects of oxygenation on
ammonia oxidation potential and microbial diversity in sediment from surface-
flow wetland mesocosms. Bioresour. Technol. 101 (4), 1389e1392.
Badhe, N., Saha, S., Biswas, R., Nandy, T., 2014. Role of algal biofilm in improving the
performance of free surface, up-flow constructed wetland. Bioresour. Technol.
169, 596e604.
Chung, J., Li, X., Rittmann, B.E., 2006. Bio-reduction of arsenate using a hydrogen-
based membrane biofilm reactor. Chemosphere 65 (1), 24e34.
Cowman, J., Torres, C., Rittmann, B.E., 2005. Total nitrogen removal in an aerobic/
anoxic membrane biofilm reactor system. Water Sci. Technol. 52, 115e120.
Drayer, A.N., Richter, S.C., 2016. Physical wetland characteristics influence
amphibian community composition differently in constructed wetlands and
natural wetlands. Ecol. Eng. 93, 166e174.
Hasar, H., Xia, S., Ahn, C.H., Rittmann, B.E., 2008. Simultaneous removal of organic
matter and nitrogen compounds by an aerobic/anoxic membrane biofilm
reactor. Water Res. 42 (15), 4109e4116.
Ifabiyi, I.P., 2008. Self purification of a freshwater stream in I1e-Ife: lessons for
water management. J. Hum. Ecol. 24, 131e137.
Lee, K.-C., Rittmann, B.E., 2002. Applying a novel autohydrogenotrophic hollow-
fiber membrane biofilm reactor for denitrification of drinking water. Water
Res. 36 (8), 2040e2052.
Moussavi, G., Jafari, S.J., Yaghmaeian, K., 2015. Enhanced biological denitrification in
the cyclic rotating bed reactor with catechol as carbon source. Bioresour.
Technol. 189, 266e272.
Ontiveros-Valencia, A., Ziv-El, M., Zhao, H.-P., Feng, L., Rittmann, B.E., Krajmalnik-
Brown, R., 2012. Interactions between nitrate-reducing and sulfate-reducing
bacteria coexisting in a hydrogen-fed biofilm. Environ. Sci. Technol. 46 (20),
11289e11298.
Ontiveros, A., Ziv-El, M.C., Zhao, H., Feng, L., Rittmann, B.E., Krajmalnik-Brown, R.,
2012. Interactions between nitrate-reducing and sulfate-reducing bacteria
coexisting in a hydrogen-fed biofilm. Environ. Sci. Technol. 40 (20),
11289e11298.
Rittmann, B.E., 2006. The membrane biofilm reactor: the natural partnership of
membranes and biofilm. Water Sci. Technol. 53, 219e226.
Rittmann, B.E., 2007. The membrane biofilm reactor is a versatile platform for water
and wastewater treatment. Environ. Eng. Res. 12, 157e175.
Rittmann, B.E., McCarty, P.L., 2001. Environmental Biotechnology: Principle and
Application. McGraw-Hil Book Co., New York.
Rittmann, B.E., Nerenberg, R., Stinson, B., Katehis, D., Leong, E., Anderson, J., 2005.
Hydrogen-based Membrane Biofilm Reactor for Wastewater Treatment. Water
Intelligence Online, No. 200504029.
Rittmann, B.E., Nerenerg, R., Lee, K.C., Najm, I., Gillogly, T.E., Lehman, G.E.,
Adham, S.S., 2004. The hydrogen-based hollow-fiber member biofilm reactor
(HFMBfR) for reducing oxidized contaminants. Water Sci. Technol. 49,
223e230.
Scherson, Y.D., Wells, G.F., Woo, S.-G., Lee, J., Park, J., Cantwell, B.J., Criddle, C.S.,
2013. Nitrogen removal with energy recovery through N2O decomposition.
Energy & Environ. Sci. 6 (1), 241e248.
Tang, Y., Zhou, C., Van Ginkel, S.W., Ontiveros-Valencia, A., Shin, J., Rittmann, B.E.,
2012. Hydrogen permeability of the hollow fibers used in H2-based membrane
biofilm reactors. J. Membr. Sci. 407e408, 176e183.
Tang, Y., Zhou, C., Ziv-El, M., Rittmann, B.E., 2011. A pH-control model for hetero-
trophic and hydrogen-based autotrophic denitrification. Water Res. 45 (1),
232e240.
Vagnetti, R., Miana, P., Fabris, M., Pavoni, B., 2003. Self-purification ability of a
resurgence stream. Chemosphere 52 (10), 1781e1795.
Wu, Y., Hu, Z., Yang, L., Graham, B., Kerr, P.G., 2011a. The removal of nutrients from
non-point source wastewater by a hybrid bioreactor. Bioresour. Technol. 102,
2419e2426.
Wu, Y., Kerr, P.G., Hu, Z., Yang, L., 2010. Removal of cyanobacterial bloom from a
biopondewetland system and the associated response of zoobenthic diversity.
Bioresour. Technol. 101 (11), 3903e3908.
Wu, Y., Xia, L., Hu, Z., Liu, S., Liu, H., Nath, B., Zhang, N., Yang, L., 2011b. The
application of zero-water discharge system in treating diffuse village waste-
water and its benefits in community afforestation. Environ. Pollut. 159,
2968e2973.
Xia, S., Li, H., Zhang, Z., Zhang, Y., Yang, X., Jia, R., Xie, K., Xu, X., 2011. Bioreduction of
para-chloronitrobenzene in drinking water using a continuous stirred
hydrogen-based hollow fiber membrane biofilm reactor. J. Hazard. Mater. 192
(2), 593e598.
Xia, S., Zhong, F., Zhang, Y., Li, H., Yang, X., 2010. Bio-reduction of nitrate from
groundwater using a hydrogen-based membrane biofilm reactor. J. Environ. Sci.
22, 257e262.
Xu, X., Xia, S., Zhou, L., Zhang, Z., Rittmann, B.E., 2015. Bioreduction of vanadium (V)
in groundwater by autohydrogentrophic bacteria: mechanisms and microor-
ganisms. J. Environ. Sci. 30, 122e128.
Yang, W., Liu, Y., Ou, C., Gabor, S., 2016. Examining water quality effects of riparian
wetland loss and restoration scenarios in a southern ontario watershed.
J. Environ. Manag. 174, 26e34.
Zhao, H., Ontiveros, A., Tang, Y., Kim, B.O., Van Ginkel, S.W., Friese, D., Rittmann, B.E.,
2014. Removal of multiple electron acceptors by pilot-scale, two-stage mem-
brane biofilm reactors. Water Res. 54, 115e122.
Zhao, H.P., Van Ginkel, S., Tang, Y., Kang, D.W., Rittmann, B.E., Krajmalnik-Brown, R.,
2011. Interactions between perchlorate and nitrate reductions in the biofilm of a
hydrogen-based membrane biofilm reactor. Environ. Sci. Technol. 45,
10155e10162.
Zhao, Z., Song, X., Wang, Y., Wang, D., Wang, S., He, Y., Ding, Y., Wang, W., Yan, D.,
Wang, J., 2016. Effects of algal ponds on vertical flow constructed wetlands
under different sewage application techniques. Ecol. Eng. 93, 120e128.
Zhou, C., Ontiveros-Valencia, A., Cornette de Saint Cyr, L., Zevin, A.S., Carey, S.E.,
 Y. Wu et al. / Water Research 119 (2017) 234e241 241

Krajmalnik-Brown, R., Rittmann, B.E., 2014. Uranium removal and microbial
community in a H2-based membrane biofilm reactor. Water Res. 64, 255e264.
Zhou, C., Ontiveros-Valencia, A., Wang, Z., Maldonado, J., Zhao, H.-P., Krajmalnik-
Brown, R., Rittmann, B.E., 2016. Palladium recovery in a H2-based membrane
biofilm reactor: formation of Pd(0) nanoparticles through enzymatic and
autocatalytic reductions. Environ. Sci. Technol. 50 (5), 2546e2555.
Ziv-El, M.C., Rittmann, B.E., 2009. Systematic evaluation of nitrate and perchlorate
bioreduction kinetics in groundwater using a hydrogen-based membrane bio-
film reactor. Water Res. 43 (1), 173e181.

View publication stats