You are on page 1of 12

Waste and Biomass Valorization

https://doi.org/10.1007/s12649-018-0223-z

ORIGINAL PAPER

Denitrification of Nitric Oxide Using Hollow Fiber Membrane


Bioreactor; Effect of Nitrate and Nitric Oxide Loadings on the Reactor
Performance and Microbiology
Vahid Razaviarani1 · Melany Ruiz‑Urigüen2 · Peter R. Jaffé2

Received: 1 November 2017 / Accepted: 29 January 2018


© Springer Science+Business Media B.V., part of Springer Nature 2018

Abstract
Nitric oxide (NO) removal from a gas stream containing ~500 ppm NO was studied in a hollow fiber membrane (HFM)
bioreactor. Compared to other biological NO removal methods the HFM bioreactor achieved NO removal rates that were as
good if not better, of up to 92% NO removal, under comparable loadings and reactor size. Results showed that a wastewater
stream containing organic carbon can be used as the electron donor to reduce the NO. Hence, combining biological NO
treatment with treatment of a wastewater containing organic carbon has may be an effective overall cost-reducing strategy.
The effect of different nitrate (NO−3 ) concentrations on the NO reduction rate was also evaluated, and results showed that
NO3− does enhance the NO removal rate. The reactor’s performance was studied under six different NO:NO−3 loading regimes
and the NO removal rate as well as the microbial denitrifier community in the reactor was tracked. Specifically, the relevant
genes responsible for each denitrification step were tracked during each different NO:NO−3 loading regime to the reactor.
Results showed that the denitrifying microbial community adjust rapidly to changes in the different N loadings, but overall
the performance of the reactor is robust and can withstand such variability in terms of NO, NO−3 and organic carbon removal.

Keywords  Nitric oxide (NO) · Hollow fiber membrane · Denitrification · Microbial community · Electron acceptor · Waste
treatment

Introduction least two times larger than the rate of the natural terrestrial
formation, and the amount of N ­ 2 produced by denitrification
The equilibrium of the nitrogen cycle has been broken due is only between 30 and 60% of the total Nr formed [3].
to human activities [1], causing several adverse environ- Nitric oxide (NO) is among the forms of Nr produced by
mental and health impacts [2]. Before the Industrial Era, human activities including power generation, industry, tran-
the sources of all reactive nitrogen species (Nr), including sit, and biomass burning [1]. Typically, over 95% of N ­ Ox in
nitrogen oxides ­(NOx) came primarily from natural sources a combustion flue gas contains NO, while nitrogen dioxide
(e.g. soil emissions and lightening), which were in balance ­(NO2) is mostly formed via photochemical reactions in the
with nitrogen (N) losses through denitrification, a process atmosphere. These flue gases, NO and N ­ O2, are two major
that returns nitrogen gas ­(N2) back to the atmosphere. By the air pollutants responsible for the formation of urban and
year 2010, the Nr formed by anthropogenic activity was at regional haze, acid rain, and ozone layer depletion, and can
contribute to climate change [4, 5]. Furthermore, exposure
to these compounds can cause severe health problems such
Vahid Razaviarani and Melany Ruiz-Urigüen contributed equally
to this work. as respiratory illness [6]. ­NOx pollution is a problem in many
urban areas. Some megacities such as Beijing, Mexico and
* Peter R. Jaffé New York City have had ­NOx concentration far beyond the
jaffe@princeton.edu
WHO standards [5], therefore, N ­ Ox pollution requires urgent
1
Department of Chemical Engineering, American University remediation and appropriate management.
of the Middle East, 250 St, Egaila, Kuwait Emissions of ­NOx have increased exponentially over the
2
Department of Civil and Environmental Engineering, past decades due to the rapid economic growth and energy
Princeton University, Princeton, NJ, USA demand from fossil fuels. This has drawn much attention

13
Vol.:(0123456789)
Waste and Biomass Valorization

globally to the enactment of stringent regulations to control NO removal is achieved when the NO and Fe(III) loading
and decrease the ­NOx emitted to the environment. Of these rates are equal to or lower than the NO and Fe(II) reduc-
oxides, ­NO2 is quite soluble in water and can be removed tion rates, respectively [14]. The deficiency of the external
easily from the off-gas stream by simple scrubbing and strip- electron donor (e.g. ethanol) causes a decline in the Fe(II)
ping methods, while NO is fairly insoluble in water and can- EDTA2− regeneration rate resulting in an increase in redox
not be separated easily [7], unless it is removed very rapidly potential. This could hamper the capacity of the NO mass
from the water phase via reaction. transfer and therefore reduces the absorption efficiencies
Conventional chemical and physical N ­ Ox controls such as for gaseous NO in the physicochemical process. Van der
selective catalytic reduction, selective non-catalytic reduc- Maas et al. [14] demonstrated that when the concentration
tion, adsorption and scrubbing are currently used to control of ethanol declined below 1 mM the NO removal efficiency
­NOx emissions. However, these technologies are cost inten- dropped dramatically from the gas phase. Van der Maas et al.
sive and produce large amounts of hazardous by-products [15], also observed that exceeding ethanol as the sole exter-
such as ammonia and particulate matter [8]. nal electron donor in the bioreactor feed lowers the redox
Biological ­NO x removal technologies have recently potential below − 300 mV and the process is outcompeted by
drawn much attention due to the lower operating costs, other biological competitors, e.g. sulfate reducing bacteria
lower energy demands and absence of secondary pollutants and methanogens. Therefore, to achieve a stable NO removal
requiring further treatment [9]. Enhanced NO removal using efficiency through the ­BioDeNOx process, a proper control
biological processes has been reported in several studies. of external electron donor supply is critical. Furthermore, it
Direct transfer of NO gas to the liquid phase followed by should be noted that the cycling of the Fe(II)EDTA2− in the
biological reduction of NO has also been studied by several ­BioDeNOx process is a closed system, hence it is exceed-
researchers. NO removal via biotrickling filter (BTF) was ingly difficult to use waste-water containing organic carbon
demonstrated by Wei et al. [10] who studied thermophilic as the electron donor.
(~50 °C) aerobic denitrification process in the removal of To overcome the limitations of the low solubility of NO
NO from gas streams. In this work, a novel aerobic denitri- in water, membrane bioreactors have been employed, and
fier Chelatococcus daeguensis was isolated from the biofilm among the different types of membrane reactors, hollow
of an on-site BTF and inoculated in a 2500 mL BTF. Results fiber membrane (HFM) bioreactors possesses a number of
showed that with the inlet NO concentration between 100 advantages over the other forms of membranes such as high
and 500 ppm the BTF system could remove NO above 80% packing density and large surface contact area [16]. Zhang
at the presence of 8% oxygen. Hiroyasu et al. [11] studied et  al. [16] treated NO from a simulated flue gas stream
NO removal in a bubble-column type of reactor seeded by through the denitrification process using a HFM bioreac-
the microalgae, Dunaliella tertiolecta. The NO removal tor. These researchers reported the maximum NO removal
percentage was about 60% in a continuous culture under efficiency of 86% for a gas stream with 1000 ppm NO and a
light condition. Jun et al. [12] used an anaerobic rotating gas flow rate of 20 mL/min. Min et al. [17] used a nitrifying
drum biofilter to examine the NO removal percentage from biofilm in a 500 mL-HFM bioreactor to remove NO from
an off-gas stream. They reported a removal efficiency of 60 combustion gases. A maximum NO removal efficiency of
and 85% when the NO concentration was varied between 67 74% was reported at the NO concentration of 100 ppm and
and 323 ppm, respectively. the flow rate of 100 mL/min. They demonstrated that the
Transfer of NO from the gas to the liquid phase can be NO removal efficiency remained almost constant regard-
enhanced significantly in the presence of Fe(II)EDTA2−. In less of the gas composition or temperature. Enhanced NO
the ­BioDeNOx process, ­NOx is absorbed by the chelated removal through the HFM bioreactor has been reported to be
Fe(II)EDTA2− in a physicochemical reactor. This step is due to the improved support for the microbial population as
fast and results in an increased mass transfer rate of ­NOx. well as the increased surface area for the NO mass transfer.
Subsequently, in the second step, biological reduction of Although, nitrification of NO using the HFM bioreactor has
­NOx to ­N2 via denitrification occurs when an adequate largely been examined in previous studies [17, 18], to the
amount of external electron donor, e.g. ethanol, is available authors knowledge, only one study has been conducted on
[13]. Though, a stable operation of the ­BioDeNOx process the NO denitrification process in these reactors [16].
is much more complex where the two microbial reactions, Denitrification of NO through various technologies has
Fe(III)EDTA− reduction and NO reduction, compete for the been reported to range from 68 to 92% removal percentage
electron donor supplied to the bioreactor. It was reported in previous studies [19–21], as shown in Table 3.
that when the redox potential of the bioreactor remained The performance and stability of the NO removal through
well below − 140 mV, the complete regeneration of Fe(II) the denitrification process depends on a series of sequential
EDTA2− as the key reaction resulted in a stable NO removal biological reactions ­(NO3−⟶NO2−⟶NO⟶N2O⟶N2)
efficiency from the process. This successful continuous catalyzed by four types of enzymes encoded by

13
Waste and Biomass Valorization

functional genes of the same name: nitrate reductase (Nar), the form of nitrate ­(NO3−-N). Higher ­NO3−-N loads might
nitrite reductase (Nir), nitric oxide reductase (Nor), and result in additional production of NO during denitrification,
nitrous oxide reductase (Nos). Furthermore, microbial stud- and hence make the reactor less efficient in its gas-phase
ies on the denitrification of NO linked to the performance NO removal rate, or vice versa, it could stimulate denitrifier
of HFM bioreactor when altering the N composition and activity and enhance the overall NO removal rate. Therefore,
concentrations have yet to be investigated. To sustain the one of the key objectives of this study was to investigate the
microbial community for denitrification, the pH in bioreac- effects of ­NO3−-N loading on the biological NO removal
tor must be monitored and controlled to remain near the through the denitrification process in a HFM bioreactor, to
optimum value of 8.0 as suggested by Niu et al. [22]. Zhang optimize the performance of such a reactor, and to determine
et al. [16] observed a significant increase in NO removal if ­NO3−-N levels are of concern in the selection of waste-
efficiency when pH increased from 6 to 8, and a decline water used for the reduction of NO. This goal was achieved
when the pH further increased from 8 to 9. The denitrifica- by altering the N­ O3−-N concentration in the feed solution
tion activity conducted outside of the ideal pH range could and the NO flow rates supplied to the membrane assembly
result in the accumulation of intermediate products includ- of the reactor, and examining the reactor’s performance and
ing nitrite and nitrous oxide [23, 24]. microbial ecology once steady state was reached.
A possible advantage of HFM-based biological NO
removal, versus a closed system like the B ­ ioDeNOx pro-
cess, is that NO can be used as an electron acceptor for the Materials and Methods
degradation of wastewater contaminated with biodegradable
organic compounds, hence resulting in energy savings dur- Bioreactor Setup and Operation
ing the wastewater treatment because of decreased needs for
aeration. Various wastewaters containing organic carbon can The denitrifying HFM bioreactor system is shown in Fig. 1.
be used as the electron donor for the denitrification process. The experiment was conducted in parallel in two separate
These wastewaters could contain varying degrees of N in bioreactors, HFM setup and a semi-continuous bioreactor

(5) (8) (1) Hollow fiber membrane


(2) NO cylinder
(3) N2 cylinder
(1)
(4) Gas flow meter
(5) NO analyzer
(6) Mixing tank (MT)
(7) (7) Pump
(8) pH controller
(9) pH probe
(6) (10) Stirrer plate
(11) Stirrer bar
(9) (12) Substrate tank

(11)

(12)
(10)

(4)

(4)

(2) (3)

Fig. 1  Schematic of HFM bioreactor system

13
Waste and Biomass Valorization

(BR), the latter used as a control reactor, both operated and the maximum NO transfer at the optimum water flow
at room temperature (25 ± 1 °C), following the different rate was determined in the absence of biomass.
N feeding protocols shown in Table 1. During the entire
experiment one bioreactor (BR) received ­NO3−-N as the Inoculum Acclimation
only source of N while the other was operated as the main
bioreactor (HFM setup, Fig. 1) and received varying loads A fresh batch of mixed culture (about 1000 mL) containing
of ­NO3−-N and NO as N sources in various stages, as shown denitrifying bacteria was collected from the effluent of a
in Table 1. A PermSelect hydrophobic polydimethylsiloxane landfill leachate treatment facility in Princeton, New Jer-
(PDMSXA-2.1) hollow-fiber membrane module (MedArray sey. The acclimation was conducted for 40 days to obtain
Inc., Ann Arbor, MI, USA) was used to transfer NO from the initial mixed liquor volatile suspended solids (MLVSS) of
gas to the aqueous phase, serving also as a support for the 4500–5000 mg/L. The nutrient composition for biomass
microbial population conducting the reduction of NO. The acclimation was prepared based on the stoichiometry reac-
module contained 30,000 fibers with an active surface area tion of heterotrophic denitrification. The HRT was kept at
of 2.1 m2. The inner and outer diameters of the fibers were 6 h to maintain the working volume and biomass concen-
0.19 and 0.3 mm, respectively. The effective shell volume tration persistent. The acclimation process was deemed
was 190 mL which was about 38% of the total bioreactor’s complete when the changes in the effluent of MLVSS and
working volume. Initially, both bioreactors were filled with ­NO3− values measurements over a 5-day period were < 5%.
500 mL of acclimated biomass and the reactor headspace At this point, the acclimated biomass was separated in half,
was purged with argon gas for 2 min. They were fed with one-half was destined as the seed for the main bioreactor,
the same feed compositions used in the acclimation phase, the HFM setup, and the second one destined as the seed
mixed thoroughly using magnetic stirrers, and were oper- for the BR. The composition of the nutrient solution fed
ated at a hydraulic retention time (HRT) of 6 h and a solids to the bioreactors was as follows: 150 mg/L ­CH3COONa,
retention time of 40 days. During the experiment, the HFM 100 mg/L ­KNO3, 17 mg/L ­KH2PO4, and 8 mg/L ­K2HPO4.
bioreactor received NO gas supplied from gas cylinder as Simulated flue gas was a mixture of NO (500 ppm in ­N2,
N source in addition to ­KNO3− present in the feed solution. v/v) and ­N2 (99.95%).
­ O3−-N and NO
During the operation, the effects of varying N
loadings were determined in six stages as shown in Table 1. Analytical Methods

Preliminary Performance of the HFM Reactor Effluent samples were collected at steady-state conditions in
each stage and were stored for subsequent analyses. Steady
The performance of the HFM in the transfer of NO from state operation was attained when coefficients of variation of
the gas to the liquid phase was first examined using a series effluent ­NO3−-N, NO, total organic carbon (TOC), and pH
of measurements in the absence of biomass. The removal were < 5%. Nitrite was measured according to the Nitrite-
percentage of NO was measured using tap water in the shell Diazotization (10019) method using a HACH DR3900
side and NO gas (500 ± 30 ppm) in the lumen side of the spectrophotometer. An Eco Physics Inc. CLD 60 series NO
HFM. At room temperature (~25 °C), the total gas flow rate analyzer was used to measure the inlet and outlet NO con-
(NO and ­N2 as makeup gas) was maintained at 1 L/min, and centrations. The inlet NO flow rate balanced with ­N2 was
the liquid flow rate varied within the acceptable ranges sug- controlled using two calibrated Cole-Parmer Aluminum/
gested for the HFM operation. The liquid circulation flow Glass flow meters. The pH in the reactors was monitored
rate was increased progressively from 0.24 to 2.15 L/min, and controlled continuously using an automatic BlueLab pH

Table 1  Bioreactors substrate Parameters BR HFM


loading characteristics
Stage
1 2 3 4 5 6

NO3−-N (g/L) 4.6 4.6 4.6 4.6 4.6 2.3 0


NOa (L/day) – 144 288 0 288 288 288
TOC (g/L) 17.8 17.8 17.8 17.8 17.8 17.8 17.8
Duration (day) 137 20 25 20 27 30 15

BR Batch (bio)Reactor, HFM hollow fiber membrane bioreactor


a
 The concentration of NO was maintained at 500 ± 30 ppm using ­N2 as the makeup gas

13
Waste and Biomass Valorization

controller. The ­NO3−-N was measured with the Nitrate-HR and denitrifying bacteria present in the samples were deter-
(10020) method using a HACH DR/3900 spectrophotom- mined by Quantitative PCR (qPCR) using specific primer
eter. For TOC and total dissolved nitrogen (TN) analyses, sets reported in the literature (Table 2), which amplify the
samples were filtered through 0.2 µm nylon filter syringes genes encoding narG, nirS, CnorB and nosZ enzymes. Fur-
and then measured using a Shimadzu TOC/TN analyzer. All thermore, to compare biomass production and sustainability
analyses were performed in triplicate. over time, a total bacteria count was conducted by amplify-
ing a region of the 16S rDNA gene using the primer set
Denitrifiers Composition Analysis 1055F/1392R. The qPCR assays were carried out using the
Applied Biosystems StepOnePlus™ Real Time PCR system.
Sludge Sampling and DNA Extraction Each qPCR mixture (20 µL) was composed of 10 µL of
SYBR Premix Ex Taq® II 2X (Takara, Japan), 0.8 µL of
A 5 mL sludge sample was collected from the acclimated each forward and reverse 10 µM primer, and DNA tem-
biomass prior to starting the reactors operations. From the plate. Thermal cycling conditions were initiated for 30 s
HFM bioreactor, 13 mL were sampled from the mixing at 95 °C, followed by 40 cycles of 5 s at 94 °C, 30 s of
tank during stages 1–6. At the end of the operation period, annealing according to primers’ specific temperature indi-
6.5 mL from the biofilm in the HFM compartment were cated in Table 2, and 30 s at 70 °C. Finally, a melting curve
obtained via destructive sampling of the HFM. From the analysis for SYBR Green assay was run to distinguish the
BR, 13 mL of sludge sample were obtained at the end of targeted PCR product from the non-targeted PCR product.
the last stage. Total genomic DNA was extracted using Fast Each qPCR reaction was run in triplicate and included nega-
DNA® Spin kit for soil (Biomedical, USA) according to tive controls and a standard curve; the last one consisting of
the manufacturer’s instructions. The sludge samples were serial dilutions of known numbers of copies of DNA of the
initially centrifuged at 10,000 rpm for 10 min and the super- gene per volume. Standards were generated via PCR using
natant was decanted carefully to obtain the pellet of biomass TaKaRa Ex Taq® (Takara, Japan) following the manufac-
for DNA extraction. All the DNA samples were preserved at tures protocol, by amplification of each targeted gene using
− 20 °C until further analysis. the primers indicated in Table 2 and total genomic DNA
from Thiobacillus denitrificans (ATCC 25259C-5) as the
Denitrifier Quantification Analysis template. The standards’ single amplicon purity and size
was confirmed by 0.8% agarose gel electrophoresis and then
Gene abundances of relevant functional genes have been purified using QIAquick® PCR Purification Kit (Qiagen).
shown to serve as a tool to predict the capacity of an eco- Qubit 2.0® (Invitrogen, USA) was used to measure DNA
system to carry out a given process such a denitrification concentration which was then used to calculate the standards
[25]. In this study we chose to detect denitrifying bacteria by concentration according to each amplicon size.
targeting relevant genes responsible for each denitrification
step: narG for nitrate reduction, nirS for nitrite reduction as
it has been shown to be the gene present in denitrifiers better
adapted to waterlogged systems [25], CnorB for NO oxida-
tion, and nosZ for ­N2O oxidation. Quantification of total

Table 2  Primer sets used to Target gene Primers Primer sequence (5′–3′) Annealing tem- Reference
target different bacteria groups perature (°C)

narG narG328F GAC​AAA​CTT​CGC​AGCGG​ 61 [26]


narG497R TCA​CCC​AGG​ACG​CTG​TTC​ 61
nirS nirS1F CCT​AYT​GGC​CGC​CRC​ART​ 57 [27]
nirS3R GCC​GCC​GTC​RTG​VAGGAA​ 57
CnorB cnorB2F GAC​AAG​NNNTAC​TGG​TGGT​ 57 [28]
cnorB7R TGNCCRTGNGCNGCNGT 57
nosZ nos1527F CGC​TGT​TCHTCG​ACA​GYCA​ 56 [29]
nos1773R ATR​TCG​ATC​ARC​TGBTCGTT​ 56
Bacterial 16s rDNA 1055F ATG​GCT​GTC​GTC​AGCT​ 55 [30]
1392R ACG​GGG​CGG​TGT​GTAC​ 55

F forward and R reverse primers

13
Waste and Biomass Valorization

RNA Extraction, RT‑qPCR and Gene Expression flow rates, as shown in Fig. 2. As indicated in literature [17,
Quantification 18], the mass transfer of poorly soluble gases trough mem-
branes is more dependent on the liquid phase turbulence
To determine shifts on narG and nirS gene transcripts due than on the gas phase turbulence, hence, increase in liquid
to ­NO3−-N loading changes (4.6, 2.3, 0 and 9.2 g/L) under velocity, or flow rate, improved the gas transfer across the
constant NO loading (288 L/day), RNA was extracted from membrane between the gas and liquid phases, as also seen
the HFM Reactor’s Mixed Tank samples obtained at the end in our results. NO removal without biomass reached a maxi-
of stages 4–6. Additionally, RNA analysis was carried out mum of 6.22%.
on a sample of the biofilm formed on the membrane fibers,
for this purpose the reactor had to be destructively sampled. HFM Bioreactor Performance
PureLink® RNA Mini Kit (Invitrogen, USA) was used to
extract RNA from 15 mL of sludge from the mixing tank. The NO removal rate in the gas stream and the N ­ O3−-N
The sludge samples were centrifuged at 10,000 rpm for removal rate in the liquid stream were monitored carefully
15 min, the supernatant was discarded and the formed pellet throughout the experiment. The experiment was conducted
was used as the substrate for the RNA extraction, accord- in six stages (Table 1). During stage 1, the concentration
ing to the manufacturer’s instructions. Immediately after the of ­NO3−-N in the feed was 4.6 g/L (corresponding to 100%
RNA extraction, cDNA of nirS and nosZ genes were pro- loading shown in Fig. 3), the gas flow rate was set to 144 L/
duced via reverse transcription using qScript™ One-Step day with a NO concentration of 518 ppm, using ­N2 as the
SYBR® Green qRT-PCR kit, ROX™ (Quanta Biosciences, makeup gas. Once steady state conditions were reached, the
USA), and corresponding primer sets indicated in Table 2. NO concentration in the gas outlet decreased to 50 ppm.
This represents a NO removal of over 90% at the gas-phase
Statistical Analysis with residence time of 2 min. At this stage, the NO and
­NO3−-N removal rates were 0.46 and 18.37 g/L day in the
To determine if the samples show a significant difference HFM bioreactor, respectively. During stage 2, the gas flow
between gene quantity and gene expression at each stage, rate was increased to 288 L/day while the inlet NO concen-
ANOVA and MANOVA-Pillai’s trace statistical analysis tration (~525 ppm) as well as the N ­ O3−-N loading remained

were run using R software (R Development Core Team, unchanged (100% N ­ O3 -N loading). Monitoring continued
2014). until steady state conditions were reached. A NO removal of
81% was observed at the gas-phase residence time of 1 min,
which corresponds to the reduction of NO concentration
Results and Discussion from 525 ppm in the inlet to 100 ppm measured in the outlet.
Therefore, during stage 2, at the constant N­ O3−-N loading
HFM Reactor Performance without Biomass (100%), doubling the NO flow rate to 288 L/day resulted in
a 76% increase in the NO removal rate relative to that during
The performance of the HFM was investigated when no bio- stage 1. This represents a NO removal rate of 0.81 g/L day
mass was present in the reactor. The experiment was con- in the HFM bioreactor. During stage 3, the N ­ O3−-N loading
ducted at constant NO flow rate, 1 L/min, and various water remained constant (100%) while no NO was introduced to
the system. As shown in Fig. 2, ­NO3−-N removal remained
almost constant indicating that the addition of NO to the
7.0 bioreactor does not influence the removal rate of ­NO3−-N.
6.0 After a 20-day period of operating without NO loading,
NO was introduced again to the HFM during stage 4 with
5.0
a flow rate of 288 L/day corresponding to a gas-phase resi-
NO removal (%)

4.0
dence time of 1 min. During this stage, NO removal rate
3.0 reached 0. 66 g/L day when the ­NO3−-N loading remained
2.0
unchanged. A NO removal efficiency of 68% was achieved
during stage 4, which is about 13% lower than during stage
1.0
2. These results indicate that although there was a slight
0.0
0 0.5 1 1.5 2 2.5
reduction in the NO removal during stage 4 compared to
Liquid flow rate (L/min) stage 2, the system had recovered significantly after no NO
was provided. This is also explained by the bacterial gene
Fig. 2  NO removal percentage at various liquid flow rates in the population dynamics which showed continuous presence
absence of biomass of all denitrifying genes throughout those stages (Fig. 4).

13
Waste and Biomass Valorization

Fig. 3  Nitrate and NO removal 40


rates at various stages. The 1.4
NO3-N (HFM) NO3-N (BR) NO (HFM)
accompanying table shows the 35
NO and ­NO3−-N loadings for 1.2

NO3-N removal rate (g/Ld)


the HFM bioreactor at each 30
stage. The BR reactor had no

NO removal rate (g/Ld)


1.0
NO input but constant ­NO3−-N 25
loading of 4.6 g/L
0.8
20

0.6
15

10 0.4

5 0.2

0 0.0
1 2 3 4 5 6
Stage

Stage 1 2 3 4 5 6
NO (L/d) 144 288 0 288 288 288
(mole/d) 2.4 4.8 0.0 4.8 4.8 4.8
NO3--N loading (%) 100 100 100 100 50 0
(mole/d) 0.66 0.66 0.66 0.66 0.33 0.0

Fig. 4  a Denitrification path- a Denitrificaon pathway and enzymes analyzed.


way, substrates and enzymes.
b Dynamics of gene families
involved in denitrification dur- narG nirS CnorB nosZ
ing the different stages of NO
and ­NO3−-N loading. Values NO3- NO2- NO N 2O N2
show the mean and SD (n = 3)

b Denitrifying genes dynamics


1.E+10

narG nirS CnorB nosZ


1.E+08
DNA copies/mL of sample

1.E+06

1.E+04

1.E+02

1.E+00
1 2 3 4 5 6
Stages

From stages 4–6 the NO flow rate remained constant at and NO loading remained unchanged. Interestingly, when
288 L/day while the ­NO3−-N loading varied. During stage no ­NO3−-N was introduced to the HFM bioreactor, the NO
5, ­NO3−-N loading was decreased by 50% compared to that removal rate was further reduced by about 23% relative to
of the previous loadings. At this loading, as shown in Fig. 3, that of stage 5, and reached 0.31 g/L day. This loading pro-
the NO removal rate was 0.40 g/L day, a reduction of 39% tocol showed the sensitivity of the denitrifier activity to the
in the NO removal efficiency compared to stage 4. During ­NO3−-N loading as indicated by the RNA quantification,
stage 6, the ­NO3−-N loading was discontinued (0% loading) further discussed in “Denitrifier Quantification Analysis”

13
Waste and Biomass Valorization

a Denitryfing genes quanficaon section. The nirS and nosZ gene transcripts showed a sig-
nificant decrease (p < 0.05) as the ­NO3−-N loading decreased
1.E+12
NarG+NirS CnorB+NosZ from 100 to 50 and 0 mg/L day uring stages 4, 5 and 6
1.E+11 respectively (Fig. 5).
Results show that NO was removed at different concen-
1.E+10 ­ O3−-N, even in the absence of nitrate,
trations of influent N
DNA copies /mL

and that increasing N ­ O3−-N loading resulted in an increase


1.E+09 in NO removal.
During the entire experiment, the concentration of
1.E+08 ­NO3−-N was well below the detection level. Several condi-
tions resulting in N­ 2O generation throughout the denitri-
1.E+07
fication process have been stated in the literature. Among
them, low pH and organic carbon deficiency are of major
1.E+06
Aclimmated MT HFM BR reasons resulting in ­N2O emissions. It has been hypothesized
seed that organic carbon scarcity results in the N ­ 2O accumula-
Inial stage Final stage tion [31], and it was revealed that the ­N2O reductase kinetic
is strongly dependent on the pH level [32]. Therefore, to
b Total 16s rRNA quanficaon minimize possible ­N2O emission in this study, the pH level
1.E+12
was kept at an optimum level of 8.0 ± 0.1, and ­CH3COONa
1.E+11
as a carbon source was continuously introduced to the
DNA copies /mL

1.E+10 bioreactors.
1.E+09 A comparison of the results from other studies to those
1.E+08 obtained from the present study is shown in Table 3. This
1.E+07 comparison shows that the membrane reactor configura-
1.E+06 tion, as applied here, combined with the use of wastewater
Aclimmated MT HFM BR as the reductant achieves among the highest NO removal
seed
efficiencies.
Inial stage Final stage

Denitrifier Quantification Analysis


Fig. 5  Number of amplified genes per mL of sludge sampled from the
acclimated biomass used as a seed for the reactor at the initial stage,
Denitrification is a sequential biological reaction catalyzed
and at the end of the operation from the mixing tank and HFM com-
partment from the HFM bioreactor and from the Batch Reactor (BR), by four types of enzymes: nitrate reductase (Nar), nitrite
the latter was run in parallel without NO loadings. a Denitrifying reductase (Nir), nitric oxide reductase (Nor) and nitrous
genes quantification, and b 16s rDNA quantification. Values show the oxide reductase (Nos) as summarized in Fig. 4a. Hence, an
mean and SD (n = 3)

Table 3  Description of previous studies of biological NO removal systems

Bioreactor T (°C) Volume (mL) NOin (ppm) Rta (min) QNOin (mL/min) REb (%) Reference

HFMb 22 ± 2 500 500 ± 30 2–1 100–200 92–81 Present study


HFMb 30 800 1000 0.5 20 86 [16]
HFMc 20 500 100 5 100 74 [17]
Jet-BioDeNOxb 21 20,000 500 – 500 92 [20]
BioDeNOxb 50 5000 100–500 – 1000 90–68 [19]
Biotrickling ­filterb 50 2500 100–500 1.92 1000 80 [10]
CABRb 50 4000 0–500 – 2000 90 [21]
Bubble ­columnd 25 4000 300 26.7 150 50–60 [11]
Rotating ­biofiltere 10–30 3000 67–323 1.44 1600 60–85 [12]
a
 Gas residence time
b
 Denitrification process
c
 Nitrification process
d
 Microalgea, Dunaliella tertiolecta, was seeded
e
 Anaerobic rotating biofilter

13
Waste and Biomass Valorization

investigation on how these genes respond when the NO and (MANOVA p < 0.001). Quantified narG showed a decrease
­NO3−-N loading to the reactors changed during each stage of over 90% from stage 4 to 5 and 55% from stage 5 to 6.
can provide additional insights on the reactor’s performance. NirS DNA quantification showed a decrease of 20% going
The dynamics of these genes during the different phases of from stage 4 to 5 and of 33% from 5 to 6. The effect of var-
the reactor’s performance are shown in Fig. 4b. iating ­NO3−-N concentrations also showed changes on the
The effects of ­NO3−-N and NO loadings on the micro- amount of CnorB and nosZ quantified DNA (MANOVA
bial community structure was studied on the samples col- p < 0.05).
lected from the mixing tank (MT) at the end of each stage Quantifying functional genes using DNA amplification is
(1–6), from the HFM compartment at the end of the opera- a useful method to determine the overall microbial composi-
tion period via destructive sampling (Fig. 1), and from the tion at different stages and in the different reactors. However,
BR, the latter only received ­NO3−-N as N source. RNA quantification is a more optimal approach to capture
Results show that all the denitrifying genes vary sig- the immediate microbial response to changes in the reac-
nificantly (MANOVA p < 0.01) between stages. However, tor’s operation. Furthermore, nirS and nosZ genes have been
the statistical analysis cannot determine the precise effect shown to be a good predictor of denitrification potential rates
of each loading regime. Therefore, we also analyze them of the systems [25]. Therefore, we quantified the amount of
separately. transcripts of nirS and nosZ genes during stages 4–6 from
During stages 1–3, the results of the specific gene DNA HFM Reactor’s Mixed Tank samples.
quantification showed that changes in NO loading while Decrease of N ­ O3−-N resulted in the decrease of nirS and
keeping the N ­ O 3−-N inflow unchanged did not signifi- nosZ gene transcripts (Fig. 7). The major change is cap-
cantly alter the dynamics of the denitrifying genes as the tured when no N ­ O3−-N is provided in the feed, nirS and
experiment progressed. Additionally, as shown in Fig. 3, nosZ mRNA decrease by 31 and 17% respectively. Statisti-
during stages 1–3, there was no change in the ­NO 3−-N cal analysis denote that ­NO3−-N loading regimes affected
removal capacity, suggesting that NO was not a competing both nirS and nosZ expression (MANOVA p < 0.1); and the
substrate over ­NO3−-N. Moreover, Fig. 4 shows that the ANOVA analysis showed equal significant response from
levels of CnorB, the gene responsible for NO reduction, both nirS and nosZ expression (p < 0.05). This indicates that
remained fairly constant during the experiment, even in the changes in ­NO3−-N loading have similar effect on the
the absence of NO. This suggests that NO reduction was activity of both genes. The results suggest that the micro-
also occurring in response to the ­NO3−-N denitrification. bial community can rapidly adjust gene transcription as a
Furthermore, one can see from the results (Fig. 6) that the response to changes in their environment, and that denitri-
higher the NO load the more TOC was removed, demon- fication of the N gas forms are driven by the NO inflow
strating our initial hypothesis that biological removal of and by the byproducts of N ­ O3− denitrification. The ratio of
NO could assist the removal of organic compounds in the nirS:nosZ during stages 4, 5, and 6 was 2.839, 2.829, and
treatment of wastewater. 2.333 respectively.
Throughout stages 4 to 6, when NO flow was kept NO was directly delivered via the gas phase to the HFM
constant and ­NO3−-N input concentrations changed from compartment through the hollow part of each membrane
4.6 to 2.3 and 0 g/L at stages 4, 5 and 6 respectively, the thread (lumen part), where it was transferred to the aque-
genes codifying for narG and nirS decreased at each stage ous phase (shell side) across the membrane. A biofilm can

Fig. 6  TOC and TN removal for


140
the different operational stages, TOC removal rate TN removal rate
analyzed from samples taken
120
from the HFM Reactor’s Mixed
Removal rate (g/Ld)

Tank 100

80

60

40

20

0
1 2 3 4 5 6
Stage

13
Waste and Biomass Valorization

Fig. 7  Number of copies of RNA response to NO3- loading changes under constant NO load
nirS and nosZ cDNA/mL of
sample, used for quantifying 2.0E+08 6.0E+07

Copies of nosZ cDNA/mL of sample


Copies of nirS cDNA/mL of sample
­ O3−-N loading
the effect of N nirS nosZ
changes, stage 4: 4.6 g/L, stage 1.6E+08 5.0E+07
5: 2.3 g/L, and stage 6: 0 g/L,
while at constant NO loading 4.0E+07
of 288 L/day. Values show the 1.2E+08
mean and SD (n = 3) 3.0E+07
8.0E+07
2.0E+07
4.0E+07 1.0E+07

0.0E+00 0.0E+00
4 5 6
Stage

The difference in loading and type of the N source intro-


duced to the HFM compartment, the HFM Reactor’s Mixed
Tank and the BR is translated to the difference in the number
of denitrifying genes detected. The acclimated biomass at
the initial stage, used as the seed for the HFM and BR, had
a slight higher amount of genes responsible for ­NO3− and
­NO2− reduction (narG + nirS = 5.53 × 1010 copies of DNA/
mL) than the number of genes responsible of NO and N ­ 2O
10
reduction (CnorB + nosZ = 1.98 × 10 copies of DNA/mL)
(Fig. 5a). At the end of the operation, the HFM had over six
times more total biomass than the initial seed and three times
more total quantified biomass than in the BR (Fig. 5b). Fur-
thermore, the HFM bioreactor showed a higher density of all
denitrifying genes compared to the initial biomass and the
BR. The HFM bioreactor had seven times more DNA copies
of CnorB and nosZ than the initial biomass, and 1.9 × 103
more than the BR. These results show a clear difference in
the microbial composition between the two types of reactors,
HFM bioreactor and BR, and between the two compartment
that make up the HFM bioreactor, in which the HFM com-
partment is where most of the biomass accumulates. It is
also at HFM where the genes responsible for the reduction to
the N gas forms are enriched. Data shows a clear connection
between the type of N.
Fig. 8  a Schematic of NO flow through HFM thread and biofilm
­ O3− concentration change over
formed over it. b Scheme of NO and N
distance
Conclusions

then develop on the wet surface of each thread, and as the Results presented here show that a HFM bioreactor can
NO diffuses away from the surface, it allows for the biofilm achieve the same NO removal efficiency as the ­BioDeNOx
to grow, reducing the NO that diffuses from the membrane process. HFM reactors in the absence of biomass are not
and utilizing the organic carbon that diffuses into the bio- capable of removing substantial amounts of NO from the gas
film from the bulk liquid. N­ O3−-N from the bulk liquid may phase. The microbial population colonizing the membrane
also diffuse into the biofilm (Fig. 8). This explains why the on the wet side of the membrane, and reducing the NO, is
microbial composition in the biofilm is very different from required to maintain a steep diffusion gradient needed for
that in the HFM Reactor’s Mixed Tank (Fig. 5a). high NO transfer across the membrane. Results without bio-
mass showed a maximum of 6.22% NO transfer from the gas

13
Waste and Biomass Valorization

to the liquid phase, however, when biomass was introduced Although we have shown here that NO can easily be
to the HFM bioreactor, the NO removal reached 92%. removed in biological reactors such as described here, a
Although higher gas loading rates to the HFM bioreactor major challenge, which was not addressed is that ­NOx gases
might result in a higher NO removal rate, the NO removal from combustion processes are emitted at high temperatures
efficiency decreases as the gas flow rate increases if the (i.e. ~200 °C). For the biological treatments the temperature
removal rate does not increase linearly. For example, over requires to be reduced to a level that makes the biological
90% NO removal was achieved in stage 1 (at a gas resi- reduction of NO feasible. Following that treatment, the gases
dence time of 2 min), corresponding to a NO removal rate of would have to be heated again (i.e. ~150 °C) to have suffi-
0.46 g/L day. By doubling the NO flow rate while maintain- cient buoyancy. This was not addressed in this project and it
ing the N­ O3−-N loading constant during stage 2, 81% NO might limit its potential application until a cost effective way
removal was achieved with a NO removal rate of 0.81 g/L of rapidly changing the temperature can be implemented.
day. Alternatively, biological reduction of NO can be applied to
Based on the results of this work, additional research gas streams with an ambient temperature, or gas streams
should be conducted to determine the optimal membrane that do not have to be reheated before being released to the
characteristics that maximizes the overall NO removal in environment.
HFM bioreactors, since a 2-min HRT for the gas phase,
that achieved 90% NO removal might be too large for many Acknowledgements  Funding for this work was provided by the Hebei
Huafeng Coking & Power Co. The seed was provided by Dr. Eugenio
applications dealing with high gas flow rates. Giraldo from NuOrganics LLC.
This study has shown that NO can be an effective elec-
tron acceptor for the biodegradation of organic carbon under
denitrifying conditions and that as the NO load increases
the organic carbon degradation also increases. Furthermore, References
results showed that the overall NO removal benefits from
higher ­NO3−-N loads, which rather than acting as a com- 1. Dentener, F., et al.: Nitrogen and sulfur deposition on regional
peting substrate with NO, stimulates denitrifier activity and and global scales: a multimodel evaluation. Global Biogeochem.
Cycles 20(4), GB4003 (2006)
enhances the overall NO removal rate.
2. US-EPA: Nitrogen oxides (NOx) control regulations. https​://
By focusing on the microbial community, this study has www3.epa.gov/regio​n1/airqu​ality​/nox.html (2015). Accessed 16
shown that denitrifiers can sustain their population under Oct 2017
different ­NO3−-N and NO loadings, including at the absence 3. Clais, P., Sabine, C.: IPCC report. Chapter 6: carbon and other
biogeochemical cycles (2013)
of NO or N ­ O3−-N, and that the HFM bioreactor’s microbial
4. Wei, Z.S., et al.: Coupling membrane catalysis and biodegrada-
community adjust rapidly to changes in the different N load- tion for nitric oxide removal in a novel hybird catalytic membrane
ings, hence having a reactor with varying loadings of NO/ biofilm reactor. Chem. Eng. J. 296, 154–161 (2016)
NO3− might influence NO removal as discussed above, but 5. Molina, M.J., Molina, L.T.: Megacities and atmospheric pollution.
J. Air Waste Manag. Assoc. 54(6), 644–680 (2004)
overall the perforce of the reactor is robust and can with-
6. Weinberger, B., et al.: The toxicology of inhaled nitric oxide.
stand such variability. Toxicol. Sci. 59, 5–16 (2001)
DNA quantification is a good method to determine the 7. Pandey, R., Chandrashekhar, B.: Physicochemical and biochemi-
overall microbial composition as it will detect any DNA pre- cal approaches for treatment of gaseous emissions containing
­NOx. Crit. Rev. Environ. Sci. Technol. 44(1), 34–96 (2014)
sent in a cell regardless of the cell’s state. Whereas, RNA
8. Li, Z., et al.: Effect of selective catalytic reduction (SCR) on
quantification is useful as it can capture the bacterial activ- fine particle emission from two coal-fired power plants in China.
ity, meaning, that it detects only live functioning bacteria. Atmos. Environ. 120, 227–233 (2015)
RNA captured the quick response in the decrease of both 9. Chen, J., et al.: Dynamic model for nitric oxide removal by a rotat-
ing drum biofilter. J. Hazard. Mater. 168(2–3), 1047–1052 (2009)
nirS and nosZ where nosZ decreased faster (as shown by
10. Wei, L., et al.: Removal of nitric oxide in a biotrickling filter under
the decreasing nirS:nosZ) due to the decreasing of N ­ O3−-N thermophilic condition using Chelatococcus daeguensis. J. Air

in the feed. However, since during stage 6 the ­NO3 -N was Waste Manag. Assoc. 62(5), 509–516 (2012)
discontinued one might have expected not to detect nirS. At 11. Hiroyasu, N., et al.: Uptake pathway and continuous removal of
nitric oxide from flue gas using microalgae. Biochem. Eng. J. 7(3),
this point it is not clear if this is because the reactor did not
241–246 (2001)
run long enough at these stages for nirS to reach steady state, 12. Jun, C., et al.: Effect of key parameters on nitric oxide removal
even though the chemical fluxes were at steady state, or tran- by anaerobic rotating drum biofilters. Environ. Technol. 29(11),
scription is not shut down completely while denitrifiers are 1241–1247 (2008)
13. Buisman, C.J.N., et al.: Process for purifying flue gas containing
active even in the absence of N ­ O3−. Overall, we show that
nitrogen oxides (1999)
the Mixing tank of the HFM bioreactor can support a stable 14. van der Maas, P., et al.: ­NOx removal from flue gas by an inte-
and active microbial population. grated physicochemical absorption and biological denitrification
process. Biotechnol. Bioeng. 90(4), 433–441 (2005)

13
Waste and Biomass Valorization

15. van der Maas, P., et al.: Enzymatic versus nonenzymatic conver- 25. Petersen, D.G., et al.: Abundance of microbial genes associated
sions during the reduction of EDTA-chelated Fe(III) in B­ ioDeNOx with nitrogen cycling as indices of biogeochemical process rates
reactors. Environ. Sci. Technol. 39(8), 2616–2623 (2005) across a vegetation gradient in Alaska. Environ. Microbiol. 14(4),
16. Zhang, X., et al.: Removal of nitric oxide from simulated flue 993–1008 (2012)
gas via denitrification in a hollow-fiber membrane bioreactor. J. 26. Reyna, L., Wunderlin, D.A., Genti-Raimondi, S.: Identification
Environ. Sci. 25(11), 2239–2246 (2013) and quantification of a novel nitrate-reducing community in sedi-
17. Min, K.-N., Ergas, S.J., Harrison, J.M.: Hollow-fiber membrane ments of Suquia River basin along a nitrate gradient. Environ.
bioreactor for nitric oxide removal. Environ. Eng. Sci. 19(6), Pollut. 158(5), 1608–1614 (2010)
575–583 (2002) 27. Braker, G., Fesefeldt, A., Witzel, K.P.: Development of PCR
18. Kumar, A., et al.: Modeling of a hollow fiber membrane biofilm primer systems for amplification of nitrite reductase genes (nirK
reactor for nitric oxide removal: Model development and experi- and nirS) to detect denitrifying bacteria in environmental samples.
mental validation. J. Chem. Technol. Biotechnol. 85(3), 423–428 Appl. Environ. Microbiol. 64(10), 3769–3775 (1998)
(2010) 28. Braker, G., Tiedje, J.M.: Nitric oxide reductase (norB) genes from
19. Li, W., Wu, C.-Z., Shi, Y.: Metal chelate absorption coupled with pure cultures and environmental samples. Appl. Environ. Micro-
microbial reduction for removal of NOx from flu gas. Chem. Tech- biol. 69(6), 3476–3483 (2003)
nol. Biotechnol. 81(3), 306–311 (2006) 29. Scala, D.J., Kerkhof, L.J.: Nitrous oxide reductase (nosZ) gene-
20. Durmazpinar, S., et al.: Biological ­NOx removal by denitrification specific PCR primers for detection of denitri¢ers and three nosZ
process in a jet-loop bioreactor: system performance and model genes from marine sediments. FEMS Microbiol. Ecol. 162, 61–68
development. Environ. Technol. 35(9–12), 1358–1366 (2014) (1998)
21. Wei, L., et al.: Biological reduction integrated system: complexed 30. Harms, G., et al.: Real-time PCR quantification of nitrifying
NO conversion pathways and nitrogen equilibrium analysis. bacteria in a municipal wastewater treatment plant. Environ. Sci.
Energy Fuels 28(7), 4725–4730 (2014) Technol. 37, 343–351 (2003)
22. Niu, H., et al.: Nitric oxide removal by wastewater bacteria in a 31. Chung, Y.-C., Chung, M.-S.: BNP test to evaluate the influence of
biotrickling filter. J. Environ. Sci. 26(3), 555–565 (2014) C/N ratio on ­N2O production in biological denitrification. Water
23. Hanaki, K., Hong, Z., Matsuo, T.: Production of nitrous oxide gas Sci. Technol. 42(3–4), 23–27 (2000)
during denitrification of wastewater. Water Sci. Technol. 26(5–6), 32. Pan, Y., et al.: Effect of pH on ­N2O reduction and accumulation
1027–1036 (1992) during denitrification by methanol utilizing denitrifiers. Water
24. Glassa, C., Silversteinb, J.: Denitrification kinetics of high nitrate Res. 46(15), 4832–4840 (2012)
concentration water: pH effect on inhibition and nitrite accumula-
tion. Water Res. 32(3), 831–839 (1998)

13

You might also like