Chemosphere 130 (2015) 8–16

Contents lists available at ScienceDirect

Chemosphere
journal homepage: www.elsevier.com/locate/chemosphere

Remediation of nitrate–nitrogen contaminated groundwater using

a pilot-scale two-layer heterotrophic–autotrophic denitrification
permeable reactive barrier with spongy iron/pine bark
Guoxin Huang a,b, Yuanying Huang c,⇑, Hongyan Hu d, Fei Liu b, Ying Zhang e, Renwei Deng a
a
China Meat Research Center, Beijing Academy of Food Sciences, Beijing 100068, China
b
Beijing Key Laboratory of Water Resources & Environmental Engineering, China University of Geosciences (Beijing), Beijing 100083, China
c
National Research Center for Geoanalysis, Beijing 100037, China
d
Hydrogeology and Engineering Geology Prospecting Institute of Heilongjiang Province, Harbin 150030, China
e
Yunnan HITECH Environmental Protection Technology Co., Ltd., Kunming 650032, China

h i g h l i g h t s

 We proposed a heterotrophic–autotrophic denitrification permeable reactive barrier.


 The barrier achieved a high NO3 –N removal efficiency (>91%) before 38 PVs.
 The packing structure over previous ones generated more H2 and CO2.
 Aerobic heterotrophic bacteria played a dominant role in oxygen depletion.
 Spongy iron ensured a robust oxygen removal in addition to form H2.

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

Article history: A novel two-layer heterotrophic–autotrophic denitrification (HAD) permeable reactive barrier (PRB) was
Received 17 October 2014 proposed for remediating nitrate–nitrogen contaminated groundwater in an oxygen rich environment,
Received in revised form 31 January 2015 which has a packing structure of an upstream pine bark layer and a downstream spongy iron and river
Accepted 7 February 2015
sand mixture layer. The HAD PRB involves biological deoxygenation, heterotrophic denitrification,
Available online 3 March 2015
hydrogenotrophic denitrification, and anaerobic Fe corrosion. Column and batch experiments were per-
Handling Editor: Chang-Ping Yu formed to: (1) investigate the NO 3 –N removal and inorganic geochemistry; (2) explore the nitrogen
transformation and removal mechanisms; (3) identify the hydrogenotrophic denitrification capacity;
Keywords: and (4) evaluate the HAD performance by comparison with other approaches. The results showed that
Nitrate–nitrogen the HAD PRB could maintain constant high NO 3 –N removal efficiency (>91%) before 38 pore volumes
Groundwater remediation (PVs) of operation (corresponding to 504 d), form little or even negative NO2 –N during the 45 PVs, and
Heterotrophic–autotrophic denitrification produce low NH+4–N after 10 PVs. Aerobic heterotrophic bacteria played a dominant role in oxygen deple-
(HAD) tion via aerobic respiration, providing more CO2 for hydrogenotrophic denitrification. The HAD PRB sig-
Permeable reactive barrier (PRB) nificantly relied on heterotrophic denitrification. Hydrogenotrophic denitrification removed 10–20% of
Pine bark
the initial NO
3 –N. Effluent total organic carbon decreased from 403.44 mg L
1
at PV 1 to 9.34 mg L1
Spongy iron
at PV 45. Packing structure had a noticeable effect on its denitrification.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction recorded in numerous aquifers in China (Zhang et al., 2013), Italy

(Sacchi et al., 2013), Mexico (Pastén-Zapata et al., 2014), Spain
Nitrate–nitrogen contamination of groundwater has become an (Boy-Roura et al., 2013), America (Su and Puls, 2007), Canada
environmental and public health problem worldwide (Della Rocca (Levallois et al., 1998), Portugal (Mendes and Ribeiro, 2010), Ger-
et al., 2007). High NO
3 –N concentration (>10 mg L
1
) has been many (Shomar et al., 2008), and Australia (Salvestrin and Hagare,
2009). It is worth mentioning that most NO 3 –N laden groundwater
also contains dissolved oxygen (DO) (Huang et al., 2012). Public
⇑ Corresponding author at: 26 Baiwanzhuang Street, Xicheng District, Beijing
health concerns arise from the potential reduction of NO 3 –N to
100037, China. Tel.: +86 1068999579; fax: +86 1068998605.
NO2 –N and the endogenous formation of N-nitroso compounds,
E-mail address: huangy1978@126.com (Y. Huang).

http://dx.doi.org/10.1016/j.chemosphere.2015.02.029
0045-6535/Ó 2015 Elsevier Ltd. All rights reserved.
 G. Huang et al. / Chemosphere 130 (2015) 8–16
which can cause methemoglobinemia, hypertension, cancers, mal-
formation, mutation, and even death (Huang et al., 2012).
Remediating NO 3 –N contaminated groundwater has been and
continues to be an area of active research. Among the available
remediation approaches (Westerhoff and James, 2003; Wang
et al., 2009; Demiral and Gündüzoğlu, 2010), biological denitrifica-
tion, falling into heterotrophic denitrification and autotrophic
denitrification, is receiving increasing attention. Heterotrophic
denitrification is widely studied and applied in the field due to high
denitrifying rate and high treatment capacity (Liu et al., 2009).
Even so, it has shown some shortcomings such as remnant organic
carbon (requiring an intensive post-treatment process) and high
cell yield (0.6–0.9 g cells g1 NO 3 –N) (Ergas and Reuss, 2001;
Shrimali and Singh, 2001). Autotrophic denitrification supported
with hydrogen, which is known as hydrogenotrophic denitrifica-
tion, is particularly promising due to clean nature of hydrogen,
low biomass yield, relatively low cost, no external carbon source,
and avoidance of secondary organic carbon contamination (Lee
and Rittmann, 2002; van Rijn et al., 2006; Moon et al., 2008). Nev-
ertheless, hydrogenotrophic denitrification is limited by inade-
quate carbon dioxide, low solubility (1.6 mg L1 at 20 °C), high
flammability, and high cost of hydrogen gas (Haugen et al., 2002;
Jha and Bose, 2005; Della Rocca et al., 2006; Ghafari et al., 2008;
Shin and Cha, 2008).
To take the advantages and overcome the disadvantages of sin-
gle heterotrophic denitrification and autotrophic denitrification, a
combined heterotrophic–autotrophic denitrification (HAD)
approach was proposed by Della Rocca et al. (2006) and further
developed by Liu et al. (2014) and Huang et al. (2012). In the
HAD processes, zero-valent iron (ZVI) via chemical reduction
removes most oxygen to rapidly create an anaerobic environment,
supporting biological denitrification. Heterotrophic denitrifiers via
heterotrophic denitrification utilize organic carbons (cotton,
methanol, and pine bark) as electron donors to biologically reduce
most NO 3 –N and meanwhile generate CO2, favoring
hydrogenotrophic denitrification. ZVI via anaerobic Fe corrosion
forms cathodic hydrogen with the aid of hydrogenase enzymes,
supporting hydrogenotrophic denitrification as well.
Hydrogenotrophic denitrifiers employ the carbon dioxide and
cathodic hydrogen to biologically reduce the remaining NO 3 –N.
Additionally, the solid materials serve as biofilm carriers. Although
the symbiotic, synergistic, and promotive effects of the mechan-
isms mentioned above in the HAD processes were encouraging,
questions arise regarding capacity of hydrogenotrophic denitrifica-
tion and performance evaluation by comparison with other
denitrification approaches.
Passive permeable reactive barrier (PRB) is considered one of
the innovative technologies widely accepted as an alternative to
the conventional pump-and-treat for sustainable in situ ground-
water remediation (Obiri-Nyarko et al., 2014) due to more cost
effectiveness and lower maintenance in the long-term (Phillips,
2009). PRBs have been successful in removing a variety of ground-
water contaminants such as NO 3 –N, petroleum hydrocarbon, and
heavy metals (Guerin et al., 2002; Schipper et al., 2005; Zhang
et al., 2012). Over the past decades, much research work has been
done on refining site characterization techniques, developing reac-
tive media (or sorbents), and improving installation and design
(Phillips, 2009). However, only a few attempts have been made
to combine multiple remediation mechanisms and optimize pack-
ing structures.
More recently, for target NO 3 –N, several column studies have
paid particular attention to HAD PRBs. For instance, Della Rocca
et al. (2006) investigated the effect factors (e.g., flow rate, influent
NO 3 –N, and retention time) using four parallel columns packed
with steel wool and cotton (R1–R4). Liu et al. (2014) determined
the spatial and temporal variations of water quality indices using
9
two parallel columns (Columns 1 and 2) packed with spongy iron
and pine bark. In the two studies, both single- and double-layer
packing structures were designed. For the single-layer structure,
the mixed reactive materials (spongy iron and pine bark) were
packed in a reactive zone; whereas for the double-layer structure,
the ZVI (steel wool or spongy iron) was placed in the upstream lay-
er and the solid carbon (cotton or pine bark) was filled in the
downstream layer. Unfortunately, the packing structure of an
upstream solid carbon layer followed by a downstream ZVI layer
has been neglected, and furthermore data on the feasibility and
effectiveness of this packing structure are rather scarce.
Here, we report the performance of a novel two-layer HAD PRB
to remediate NO 3 –N contaminated groundwater in an aerobic
environment and further develop HAD processes and PRB systems.
Our HAD PRB concept, mainly involving a combination of biological
deoxygenation, heterotrophic denitrification, hydrogenotrophic
denitrification, and anaerobic Fe corrosion, was composed of an
upstream pine bark layer and a downstream spongy iron and river
sand mixture layer. The packing structure over the previous pack-
ing ones is expected to allow the generation of more cathodic
hydrogen and carbon dioxide to support hydrogenotrophic denitri-
fication. Heterotrophic denitrification contributed to the removal
of most NO 3 –N. Aerobic heterotrophic bacteria played a dominant
role in oxygen depletion. Spongy iron potentially ensures a robust
oxygen removal in case of temporary insufficient biological deoxy-
genation. The objectives of this study were to: (1) investigate the
NO 3 –N removal and inorganic geochemistry caused by the pro-
posed HAD PRB; (2) explore the nitrogen transformation and clean
up mechanisms; (3) identify the hydrogenotrophic denitrification
capacity in the HAD PRB; and (4) evaluate the HAD performance
by comparison with other approaches.
2. Materials and methods
2.1. Materials and chemicals
Spongy iron (0.15–2.00 mm, Brunauer–Emmett–Teller area
0.49 m2 g1, micropore volume 0.0012 cm3 g1, micropore dia-
meter 101.87 Å) was obtained from Kaibiyuan Co., Beijing, China,
which consisted of Fe0 (60.60 wt%), C (0.78 wt%), S (0.06 wt%), P
(0.04 wt%), Mn (0.29 wt%), Ni (0.02 wt%), Cr (0.02 wt%), Cu
(0.02 wt%), and Al (0.26 wt%). Pine bark (2–11 mm, Brunauer–Em-
mett–Teller area 0.46 m2 g1, micropore volume 0.0018 cm3 g1,
micropore diameter 159.42 Å) was purchased from a local nursery
store in Beijing, China. Gravel (2–5 mm) and river sand (0.45–
2 mm) were donated by a quarry in Beijing, China. Hydrogen gas
(industrial grade, purity 99.9%) was purchased from Beijing Hua
Yuan Gas Chemical Industry Co., Ltd., Beijing, China. Deionized
water was used to prepare reagent solutions. Beijing tap water
was spiked with NaNO3, NaHCO3, and K2HPO4 acting as synthetic
groundwater. In addition to NO 3 –N (20–105 mg L
1
), the synthetic
groundwater contained the following (final concentration in
+
mg L1 except pH): NO 2 –N (<0.1), NH4–N (<0.8), F

(0.3), Cl
2
(20.8), HCO 3 (254.2), SO4 (46.4), Na+ (142.3–281.9), K+ (11.7),
Ca2+ (48.5), Mg2+ (28.1), DO (5–9), P (3.0), and pH (7.5–8.5). Unless
otherwise indicated, all chemicals used were analytical reagent
grade as received.
2.2. Column experiment setup and operation
A pilot-scale flow through HAD PRB experiment was conducted
employing a plexiglas column (20.6 cm i.d., 150 cm in length)
(Fig. 1a and b). A lower support layer of 5 cm of gravel was packed
at the bottom of the column, and a 5 cm gravel cap was placed at
the top. The reactive zone was composed of an upstream layer
10 G. Huang et al. / Chemosphere 130 (2015) 8–16
(a) Effluent
Legend:
150cm
Gravel
Spongy iron
125cm
Sand
Pine bark 105cm
Water seal
Intermediate sampling ports
65cm
25cm
0cm
Influent Reservoir
Feed pump
Fig. 1. (a) Schematic and (b) photograph of the continuous flow setup in the column experiment, and (c) schematic of an incubation bottle in the batch tests.
(110 cm long) packed with pine bark (7.13 kg) and a downstream
layer (30 cm long) filled with a mixture of spongy iron (7.99 kg)
and river sand (7.40 kg), providing a pore volume (PV) of 24.77 L
and a porosity of 54.28%. The river sand was used in an effort to
assure optimal permeability, minimize potential clogging and
avoid cementation during operation. Fourteen intermediate sam-
pling ports were positioned along the column height, at 15, 25,
35, 45, 55, 65, 75, 85, 95, 105, 115, 125, 135, and 145 cm from
the influent end.
It is important to note that the inoculum accounting for 45% of
the column pore volume was introduced into the column, and then
recycled for 20 d for the microorganisms to attach onto the surface
of the media particles. The inoculum was enriched from a sub-sur-
face soil taken from a pristine and humic acid rich area, and found
to contain heterotrophic denitrifiers (belonging to Bacillus, Clostri-
dium, Flavobacterium, Steroidobacter, and Novosphingobium),
hydrogenotrophic denitrifiers (belonging to Pseudomonas), and
aerobic heterotrophic bacteria (belonging to Adhaeribacter and
Flavisolibacter) (Liu et al., 2014).
Under the steady state, the column was loaded continuously
with synthetic groundwater in an upflow mode using an adjustable
multiport peristaltic pump set (BT 100-1F drive, DG-4 pump head,
Baoding Longer Precision Pump Co., Ltd., Baoding, China). The col-
umn was covered with aluminum foil to shade out photosynthesiz-
ers (Rust et al., 2002), and operated at room temperature
(23 ± 5 °C) in 9 phases (Table S1 in Supporting Information) each
for 5 PVs, in sequence. Water samples were collected from the
influent, effluent, and intermediate sampling ports, and stored at
4 °C prior to analysis for pH, DO, total organic carbon (TOC),
+ respectively. DO and water temperature were monitored using a
NO 
3 –N, NO2 –N, NH4–N, and water temperature.
2.3. Batch test design
Batch tests were conducted in sterile rubber capped, 1 L brown
glass bottles to identify the hydrogenotrophic denitrification capa-
city (Fig. 1c). The procedures of a batch incubation were introduced
(b)
(c)
Gas tubing 1 Gas tubing 2
H2
Mixed
bacteria
+
Synthetic
groundwater
as follows: (1) add a 0.45 lm filter membrane covered with mixed
bacteria and 500 mL tap water enriched with 22 mg L1 NO 3 –N,
3.0 mg L1 K2HPO4-P, and 350 mg L1 NaHCO3 into an incubation
bottle; (2) sparge the bottle with N2 through Gas tubing 2 for
5 min at 200 kPa (Mousavi et al., 2012); (3) sparge the bottle with
H2 through Gas tubing 1 for 30 and 10 min at 50 kPa before and
after Gas tubing 2 was sealed; (4) cover the bottle with aluminum
foil; (5) statically incubate the bottle in the dark at 23 ± 5 °C; (6)
draw water samples after 3, 6, 18, 22, and 30 d; and (7) after sam-
pling, readd H2 into the bottle.
Two mixed bacterial suspensions (each 200 mL) were respec-
tively collected at 65 and 125 cm of the column after 40 PVs. To
separate the mixed bacteria from TOC and inhibit heterotrophic
denitrifying bacteria, each bacterial suspension was filtered
through a 0.45 lm filter membrane with depleting suspended solid,
and then the filtrate was passed through a 0.2 lm filter membrane
with retaining the mixed bacteria on the filter membrane.
2.4. Analytical methods and instruments
+
NO 
3 –N, NO2 –N, and NH4–N were determined using a Hewlett
Packard 8453 ultraviolet–visible spectrophotometer (USA),
employing ultraviolet spectrophotometric method at 220 and
275 nm, N-(1-naphthyl)-ethylenediamine dihydrochloride colori-
metric method at 540 nm, and Nessler’s reagent colorimetric
method at 410 nm, respectively. The lower detection limits for
+
NO 
3 –N, NO2 –N, and NH4–N were 0.08, 0.003, and 0.025 mg N L
1
,
Hach HQ30d DO portable meter (USA). pH was measured using a
Sartorius PB-10 digital pH meter (Germany). TOC was measured
using a Shimadzu 3201 TOC analyzer (Japan) by measuring the dif-
ference between the total carbon (burning at 680 °C) and inorganic
carbon (acidification with 2 M HCl). In each analysis, at least one in
five samples was duplicated and the deviation between the two
samples was always less than 5%.
 G. Huang et al. / Chemosphere 130 (2015) 8–16 11

2.5. Data processing
Analytical data are presented as means of triplicate measure-
ments. The number of PVs was expressed as the ratio of accumulated
water volume (L) over time to column pore volume (L). Total nitrogen
(TN) (mg L1) was defined as the sum of NO 1 
3 –N (mg L ), NO2 –N
1 + 1 1
(mg L ), and NH4–N (mg L ). N removed (mg L ) was defined as
influent concentration (mg L1)–effluent concentration (mg L1)
(for the column experiment) or initial concentration (mg L1)–final
concentration (mg L1) at time t (d) (for the batch tests). N formed
(mg L1) was defined as effluent concentration (mg L1)–influent
concentration (mg L1) (for the column experiment) or final concen-
tration (mg L1)–initial concentration (mg L1) at time t (d) (for the
batch tests). Removal efficiency (%) was calculated as [influent con-
centration (mg L1)  effluent concentration (mg L1)]/influent con-
centration (mg L1)  100% (for the column experiment) or [initial
concentration (mg L1)  final concentration (mg L1) at time t (d)]/
initial concentration (mg L1)  100% (for the batch tests). Denitrifi-
cation rate (g m3 d1) was calculated as flow rate (m d1)  effective
cross-sectional flow area of the column (m2)  [influent NO 3 –N
3
(mg L1)  effluent NO 1
3 –N (mg L )]/effective column volume (m ).
3. Results and discussion
3.1. Nitrate–nitrogen removal
Fig. 2 shows the temporal changes in NO 3 –N in samples from
the influent and effluent together with the results of its removal
efficiency.
As shown in Fig. 2, high NO3 –N removal efficiency (>91%) was
maintained relatively constant, and low effluent NO 3 –N
(<8.90 mg L1) remained relatively stable before 38 PVs (corre-
sponding to 504 d) of operation regardless of influent NO 3 –N and
flow rate. In addition, NO3 –N removal efficiencies were close to
97–100% at a VNLR of 65.92 g m3 d1 (Fig. S1 of Supporting Infor-
mation). These results indicated that the HAD PRB was effective
and efficient in NO 3 –N removal. But removal efficiency sharply
ating conditions during the 36–40 PVs (Fig. 2). The drop also
demonstrated the reduction of both pine bark carbon releasing
capacity and organic carbon available to the heterotrophic denitri-
fying population (Fig. 5b). It is evident that a time-dependent
decay in the HAD PRB denitrification performance was noticeable
after 38 PVs. Volokita et al. (1996a) have also reached a similar
conclusion when newspaper served as a sole carbon source and
electron donor. Afterward, removal efficiency (77–82%) tended to
reach a steady state and effluent NO 3 –N was kept at a level of
18.23–23.66 mg L1 during the 40–45 PVs (Fig. 2), which was
around its limitation value (20 mg L1 for drinking use) of quality
standard for ground water of China (State Bureau of Technical
Supervision (SBTS), 1993).
3.2. Nitrogen transformation
Fig. 3 shows the temporal changes in nitrogen.
As was apparent from Fig. 3, TN removed was significantly low-
er than NO 3 –N removed at the beginning of operation (up to 10
PVs), which was attributed to the high NH+4–N formation. After
10 PVs, TN removed was nearly equal to NO 3 –N removed, which
was a proof that relatively low NH+4–N and NO 2 –N were generated
(Fig. 3). Little NO2 –N formation of <0.05 mg L
1
or even temporary
negative formation of <0.02 mg L1 (absolute value) was found
during operation over the 45 PVs (Fig. 3). This suggested that the
activities of nitrite reductase were higher than or equal to those
of nitrate reductase, and nitrite reductase quickly adapted itself
to a new environment and refreshed its activity and synthesis
when operating conditions were changed. No NO 2 –N accumula-
tion is a desirable attribute because of reducing the risk to public
health. High NH+4–N formation with an average of 4.37 mg L1
was noted during the first 10 PVs mainly due to the wash-out of
some ammonium fertilizer applied to the pine bark beforehand,
and subsequently low formation of <1.90 mg L1 was observed
mainly due to the deamination of organic nitrogen from the pine
bark (Fig. 3) (DeSimone and Howes, 1998; Liu et al., 2014). No mat-
ter what the pathways are for the NH+4–N generation, the sum of
NO +
2 –N and NH4–N formed was much less than NO3 –N reduced


dropped down to 82% and effluent NO 3 –N drastically rose up to at all times (Fig. 3), indicating most of the influent NO 3 –N was
18.23 mg L1 at PV 39 (Fig. 2). The drop in removal efficiency biologically transformed into gaseous nitrogen, as expressed by
demonstrated that influent NO 3 –N (103 mg L
1
) and flow rate Eq. (1) (Fernández-Nava et al., 2010).
1
(0.22 m d ) were not the causes of the HAD PRB performance
NO3 ! NO2 ! NOðgÞ ! N2 OðgÞ ! N2 ðgÞ ð1Þ
decrease, because the column was operated under the same oper-

Influent NO3--N 23 mg L -1 48 mg L-1 103 mg L-1

Flowrate 0.15 m d-1 0.22 m d-1 0.30 m d-1 0.15 m d-1 0.22 m d-1 0.30 m d-1 0.15 m d-1 0.22 m d-1 0.30 m d-1
Phase ĉ Ċċ Č č Ď ď Đ đ
120 100
110 90
100
NO3--N percent removal (%)

80
90 In Out Removal efficiency 70
NO3--N (mg L-1)

80
70 60
60 50
50 40
40
Standard for drinkinguse (China) 30
30
20
20
10 10
0 0
0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45
Number of pore volumes

Fig. 2. Changes in NO

3 –N in the influent and effluent and its removal efficiency over time under variable operating conditions. Vertical dashed lines represent the times when
influent NO 3 –N and/or flow rate was changed.
12 G. Huang et al. / Chemosphere 130 (2015) 8–16

120
0.06
108
0.04
96 0.02
0.00
84
-0.02 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45
72 -0.04
N (mg L-1)

NH4+-N formed
60 NO2--N formed

NO3--N removed
48
TN removed
10
36 8
6
24 4
2
12 0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45
-12
Number of pore volumes

Fig. 3. Changes in nitrogen over time. Total nitrogen (TN) (mg L1) = NO
3 –N (mg L
1
) + NO 2 –N (mg L
1
) + NH+4–N (mg L1). NO3 –N or TN removed (mg L
1
) = influent
+
concentration (mg L1)  effluent concentration (mg L1). NO
2 –N or NH4–N formed (mg L
1
) = effluent concentration (mg L1)  influent concentration (mg L1).

(a) 150 150

Influent NO3--N of 23 mg L-1 (b)
Influent NO3--N of 46 mg L-1
125 Influent NO3--N of 103 mg L-1 125
Column height (cm)
Column height (cm)

100 100

75 75

50 50

25 25

0 0
0 12 24 36 48 60 72 84 96 108 120 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 4.0
NO3--N (mg L-1) NO2--N (mg L-1)

150 150
(c) (d)
125 125
Column height (cm)

Column height (cm)

100 100

75 75

50 50

25 25

0 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 0 1 2 3 4 5 6 7 8 9 10
NH4+-N (mg L-1) DO (mg L-1)
+
Fig. 4. Changes in (a) NO  
3 –N, (b) NO2 –N, (c) NH4–N, and (d) dissolved oxygen (DO) along the column at three influent NO3 –N concentrations of 23 mg L
1
(pore volume (PV)
11), 46 mg L1 (PV 26), and 103 mg L1 (PV 41). Arrows indicate flow direction.

3.3. Spatial nitrogen distribution The changes in NO 3 –N at the influent concentrations of 23 and
46 mg L1 were similar to each other upward through the column:
Fig. 4 shows the spatial changes in nitrogen and oxygen at three almost all the introduced NO 3 –N (P97%) was depleted within the
influent NO 3 –N concentrations of 23 mg L
1
(PV 11), 46 mg L1 (PV first 65 cm; and the resultant low NO 3 –N (<1.50 mg L
1
)
1
26), and 103 mg L (PV 41). approached to a steady state between 65 and 150 cm (Fig. 4a). In
 G. Huang et al. / Chemosphere 130 (2015) 8–16 13

(a)
Influent NO3--N 23 mg L-1 48 mg L-1 103 mg L-1

Flowrate 0.15 m d-1 0.22 m d-1 0.30 m d-1 0.15 m d-1 0.22 m d-1 0.30 m d-1 0.15 m d-1 0.22 m d-1 0.30 m d-1
Phase ĉ Ċ ċ Č č Ď ď Đ đ
10 100
9 90

NO3--N percent removal (%)

8 80
7 70
DO (mg L-1)

6 60
5 50
4 In Out Removal efficiency 40
3 30
2 20
1 10
0 0

(b) 500
450 5.0
400
4.0
350
3.0
TOC (mg L-1)

300
2.0
250
1.0
200
0.0
150 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45
100
50
0

(c) 10.0
9.5
Standard for drinkinguse (China)
9.0
8.5
8.0
pH

7.5
7.0
6.5
6.0
Standard for drinkinguse (China)
5.5
5.0
0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45
Number of pore volumes

Fig. 5. Changes in (a) dissolved oxygen (DO), (b) total organic carbon (TOC), and (c) pH in the influent and effluent over time under variable operating conditions. Vertical
dashed lines represent the times when influent NO 3 –N and/or flow rate was changed.

the case of the influent concentration of 103 mg L1, 79% of the maximum of 2.14 mg L1 was detected in the column at the three
influent NO 3 –N was eliminated throughout the column with a high influent concentrations (Fig. 4c). In the spongy iron layer, NO
3 –N
NO 3 –N of 21.44 mg L
1
at 125 cm (Fig. 4a), suggesting that the was not chemically reduced to NH+4–N by the spongy iron, as evi-
denitrifying bacteria present in the column did not have the capa- denced by a small difference in NH+4–N at 125 cm (1.95 mg L1)
city to completely degrade high-concentration NO 3 –N due to a lack and 150 cm (1.88 mg L1) at the influent concentration of
of usable TOC. No NO 2 –N accumulation occurred along the column 103 mg L1 (Fig. 4a and c). Apparently, spongy iron-based chemical
at the influent concentrations of 23 and 46 mg L1 (Fig. 4b). Howev- reduction was not a cause of NH+4–N formation.
er, NO 2 –N sharply accumulated with a peak of 1.21 mg L
1
up to DO quickly decreased in the first 25 cm and afterward slowly
25 cm, and rapidly descended thereafter to less than 0.10 mg L1 declined to lower than 0.75 mg L1 between 25 and 130 cm at
at the influent concentration of 103 mg L1 (Fig. 4b). Contrary to any influent concentration (Fig. 4d). The sharp drop in DO around
our previous batch studies where no NH+4–N was formed when a the influent confirmed that the aerobic heterotrophic bacteria
mixture of pine bark, spongy iron, and mixed bacteria was kept in played a dominant role in oxygen depletion via aerobic respiration
incubation continuously for 105 d (Liu et al., 2014), NH+4–N with a (Eq. (2)) (Della Rocca et al., 2005). In another HAD PRB column,
14 G. Huang et al. / Chemosphere 130 (2015) 8–16

(a) 5.0 100 (b) 5.0 100

NH4+-N formed NO2--N formed
4.5 90 4.5 90
NO3--N removed TN removed
Removal efficiency 4.0 80

NO 3-- N percent removal (%)

4.0 80

NO3--N percent removal (%)

3.5 70 3.5 70

3.0 60 3.0 60

N (mg L -1)
N (mg L-1)

2.5 50 2.5 50

2.0 40 2.0 40

1.5 30 1.5 30

1.0 20 1.0 20

0.5 10 0.5 10

0.0 0 0.0 0
3 6 18 22 30 3 6 18 22 30
Reaction time (d) Reaction time (d)

Fig. 6. Changes in nitrogen and NO3 –N removal efficiency over time by hydrogenotrophic denitrification with the mixed bacteria from (a) 65 and (b) 125 cm of the column,
respectively. Total nitrogen (TN) (mg L1) = NO 3 –N (mg L
1
) + NO
2 –N (mg L
1
) + NH+4–N (mg L1). NO
3 –N or TN removed (mg L
1
) = final concentration (mg L1)  initial
+
1
concentration (mg L ) at time t (d). NO 2 –N or NH4–N formed (mg L
1
) = initial concentration (mg L1)  final concentration (mg L1) at time t (d).

Table 1
Comparison of biological denitrification capacity among different approaches.

Denitrification approach Packing material System description Maximum Maximum References

removal denitrification rate
efficiency (%) (g m3 d1)
HAD Pine bark/spongy iron/sand Double-layer PRB 99c 9.5c This study
HAD Spongy iron/sand/pine bark Media mixed single-layer 99b,c 11.7b,c Liu et al. (2014)
(Column 1) PRB
HAD Spongy iron/sand/pine bark Double-layer PRB 90b,c 8.7b,c Liu et al. (2014)
(Column 2)
Heterotrophic denitrification Iron powder/sodium citrate/ Double-layer PRB 100b 7.5b Liu et al. (2013)
activated
carbon
HAD Cotton/steel wool (R4) Double-layer column 100b 277 Della Rocca et al. (2006)
reactor
a
Heterotrophic denitrification Sawdust/soil Field pilot-scale single-layer NA 1.2 Schipper and Vojvodic-
PRB Vukovic (2000)
Heterotrophic denitrification Sawdust/sand 15-year-old pilot-scale 100b 6 Robertson et al. (2008)
single-layer PRB
Heterotrophic denitrification Cotton Field pilot-scale single-layer 100 360 Soares et al. (2000)
plant
Hydrogenotrophic Steel wool Single-layer column reactor 18%b 5.0b Till et al. (1998)
denitrification
a
No data available.
b
Estimated or calculated based on data given in the article.
c
Estimated or calculated based on NO3 –N transformed into gaseous nitrogen.

spongy iron dominantly contributed to oxygen removal via chemi-
cal reduction (Liu et al., 2014). Clearly, more CO2 was produced
during deoxygenation in this study, then providing more inorganic
carbon for hydrogenotrophic denitrifiers (Eq. (3)) (Lee and
Rittmann, 2002). It should be noted that CO2 generated in
heterotrophic denitrification (Eq. (4)) (Della Rocca et al., 2005)
was also an inorganic carbon source.
aerobic heterotrophs
2C6 H10 O2 þ 15O2 ! 12CO2 þ 10H2 O
By combining the results of Fig. 4a and d, it is therefore conclud-
ed that NO3 –N depletion and DO removal were two parallel pro-
cesses, that is, the presence of oxygen did not limit biological
denitrification. An explanation is that the dense and thick biofilms
around the pine bark near the influent as well as their stratified
structures protected the denitrifying bacteria in the inner zone of
the biofilms from oxygen inhibition.
ð2Þ 3.4. Inorganic geochemistry

NO3 þ 3:03H2 þ Hþ þ 0:229CO2
hydrogenotrophic denitrifiers
 !
þ 3:37H2 O
2C6 H10 O2 þ 6NO3 þ 6Hþ
Fig. 5 shows the temporal changes in DO, TOC, and pH in sam-
ples from the influent and effluent.
0:0458C5 H7 O2 N þ 0:477N2
It can be seen from the results that during operation over the 45
ð3Þ PVs, influent DO fluctuated between 5.84 and 8.57 mg L1, whereas
the corresponding effluent value was in the range of 0.19–
heterotrophic denitrifiers
! 6CO2 þ 10H2 O þ 3N2 1.63 mg L1, indicating the good DO removal efficiency (74–97%
removed) of the HAD PRB in any phase (Fig. 5a). The low effluent
ð4Þ
DO level is thought to be desirable, as high DO may not only cause
 G. Huang et al. / Chemosphere 130 (2015) 8–16
high biomass buildup, but may also render NO 3 –N regeneration if
NH+4–N is subsequently encountered in the downstream aquifer.
Despite the fact that the spongy iron did not function efficiently
in capturing oxygen (Fig. 4d), it was pretty essential to ensure a
robust oxygen removal in case of temporary insufficient biological
deoxygenation in the upstream pine bark layer (e.g., as a result of
increases in toxicity levels and change in seasonal water
temperature).
Effluent TOC decreased rapidly from 403.44 mg L1 at PV 1 to
108.46 mg L1 at PV 8, and then declined slowly to 9.34 mg L1
at PV 45 at the influent TOC of 62.44 mg L1 (Fig. 5b), which was
consistent with the literature describing wheat straw-supported
heterotrophic denitrification (Aslan and Türkman, 2004). The rapid
decrease during the early operational stage could be explained due
to the rapid decrease in easily soluble, amorphous components of
the cellulose and hemicellulose contained in the pine bark. Organic
carbon excess was also found in other studies (Volokita et al.,
1996b; Aslan and Türkman, 2004; Della Rocca et al., 2006; Liu
et al., 2014). Consequently, how to realize a consistent, slow, and
stable carbon release of cellulose-based solid substrates is worthy
of further study.
Effluent pH was 7.07–8.50 at any time point at the influent val-
ue of 7.69–8.37 (Fig. 5c). Accordingly, the effluent pH met the
water quality standard (6.5–8.5 for drinking use) (SBTS, 1993).
The optimum pH for biological denitrification should be kept in
the range of 7.0–9.0 (Tang et al., 2011; Liu et al., 2013). Undoubt-
edly, the pH in this study was not a cause of lowering the HAD
PRB denitrification performance after 36 PVs (Fig. 2).
3.5. Hydrogenotrophic denitrification capacity
Fig. 6 shows the temporal changes in nitrogen and NO 3 –N
removal efficiency in incubation bottles by hydrogenotrophic
denitrification supported by the mixed bacteria from 65 and
125 cm of the column, respectively.
NO 3 –N removal efficiency of 10–20% was achieved (Fig. 6a and
b), indicating that hydrogenotrophic denitrification contributed to
NO 3 –N removal in the incubation bottles. Besides, considering
those hydrogenotrophic denitrifying bacteria on the microbores
of the pine bark and spongy iron media, the hydrogenotrophic
denitrification capacity may be further strengthened. Combining
Figs. 2 and 6, hydrogenotrophic denitrification played a minor role
in NO 3 –N depletion in the HAD PRB compared with heterotrophic
denitrification. However, hydrogenotrophic denitrification was
able to minimize potential gas and biomass clogging, as it con-
sumed cathodic hydrogen formed by anaerobic ZVI corrosion and
carbon dioxide formed by heterotrophic denitrification (Eq. (3)),
and produced a low biomass yield of 0.24 g cells g1 NO 3 –N, which
was less than 0.6–0.9 g cells g1 NO 3 –N typically reported for
heterotrophic denitrification (Ergas and Reuss, 2001).
Hydrogenotrophic denitrification supported with the mixed bacte-
ria from 65 cm showed a higher removal efficiency (17–20%) than
that from 125 cm (10–15%) (Fig. 6a and b). This phenomenon
demonstrated that the hydrogenotrophic denitrifiers developed
better in the upstream pine bark layer owing to sufficient inorganic
carbon (HCO 
3 + CO2) and NO3 –N substrates. TN removed appeared
invariably a little lower than NO 3 –N removed as a result of the lit-
tle formation of NO 2 –N (60.17 mg L
1
) and NH+4–N (60.25 mg L1)
(Fig. 6a and b).
3.6. Comparison of biological denitrification performance among
different approaches
Based on a literature survey, a comparison was made of denitri-
fication performance among different approaches (Table 1).
15
Even though any approach that had heterotrophic denitrifica-
tion included was capable of attaining the ideal NO 3 –N removal
efficiency (P90%), the cotton supported approach presented a
much higher denitrification rate than the other ones (including
the HAD approach in this study) (Table 1). Nevertheless, the soft
texture of cotton limits its in situ applicability on a large scale in
that it will be easily and strongly compressed at a high flow rate
(Soares et al., 2000). The single steel wool mediated
hydrogenotrophic denitrification was not able to efficiently
remove the influent NO 3 –N (18% removed) (Table 1). The differ-
ences in denitrification performance probably resulted from the
differences in hydraulic retention time, diversity of the used inocu-
lum, denitrification mechanism, and so on. In spite of the similar
packing media (spongy iron/pine bark/sand) and denitrification
mechanisms, the HADs achieved different removal efficiencies
and denitrification rates (Table 1), demonstrating that packing
structure had a noticeable effect on HAD PRB denitrification.
4. Conclusions
Based on the research findings the main conclusions were as
follows:
1. The proposed HAD PRB was effective and efficient in NO 3 –N
removal. High NO 3 –N removal efficiency (>91%) was main-
tained relatively constant before 38 PVs (corresponding to
504 d). Hydrogenotrophic denitrification (removal efficiency
10–20%) played a minor role in NO 3 –N depletion compared
with heterotrophic denitrification.
2. Most influent NO 3 –N was biologically transformed into gaseous
nitrogen. Little NO 1
2 –N formation of <0.05 mg L or even tem-
1
porary negative formation of <0.02 mg L (absolute value) was
found during the 45 PVs. Low NH+4–N formation of <1.90 mg L1
was observed after 10 PVs caused by the deamination of organic
nitrogen from the pine bark rather than the spongy-iron based
chemical reduction.
3. Almost all the introduced NO 3 –N (P97%) was depleted within
the first 65 cm of the column at the influent NO 3 –N concentra-
tions of 23 and 46 mg L1, whereas 79% of the influent NO 3 –N
was eliminated throughout the column at the influent concen-
tration of 103 mg L1.
4. Aerobic heterotrophic bacteria played a dominant role in oxy-
gen depletion via aerobic respiration, providing more CO2 for
hydrogenotrophic denitrification. Spongy iron was pretty
essential to ensure a robust oxygen removal.
5. Effluent pH (7.07–8.50) met the water quality standard (6.5–8.5
for drinking use). Effluent DO (0.19–1.63 mg L1) was low and
desirable. The rapid decrease in effluent TOC from 403.44 to
108.46 mg L1 during the first 8 PVs was due to the rapid
decrease in easily soluble, amorphous components of the cellu-
lose and hemicellulose.
6. Even though the packing media and denitrification mechanisms
are similar in the HAD PRBs, packing structure had a noticeable
effect on their denitrification performance.
Acknowledgements
This study is financially supported jointly by NSF – China
(41172226), National Program of Control and Treatment of Water
Pollution (2009ZX07424-002-002), China Postdoctoral Science
Foundation – China (2013M541111) and Beijing Excellent Talents
Program (2012D001055000001).
16 G. Huang et al. / Chemosphere 130 (2015) 8–16
Appendix A. Supplementary material
Table S1 provides the operating conditions of the column in the
9 phases. Fig. S1 shows the changes in removal efficiency over
volumetric NO 3 –N loading rate. Supplementary data associated
with this article can be found, in the online version, at http://
dx.doi.org/10.1016/j.chemosphere.2015.02.029.
References
Aslan, S., Türkman, A., 2004. Biological denitrification of drinking water using
various natural organic solid substrates. Water Sci. Technol. 48, 489–495.
Boy-Roura, M., Menció, A., Mas-Pla, J., 2013. Temporal analysis of spring water data
to assess nitrate inputs to groundwater in an agricultural area (Osona, NE
Spain). Sci. Total Environ. 452–453, 433–445.
Della Rocca, C., Belgiorno, V., Meric, S., 2005. Cotton-supported heterotrophic
denitrification of nitrate-rich drinking water with a sand filtration post-
treatment. Water SA 31, 229–236.
Della Rocca, C., Belgiorno, V., Meric, S., 2006. An heterotrophic/autotrophic
denitrification (HAD) approach for nitrate removal from drinking water.
Process Biochem. 41, 1022–1028.
Della Rocca, C., Belgiorno, V., Meriç, S., 2007. Overview of in-situ applicable nitrate
removal processes. Desalination 204, 46–62.
Demiral, H., Gündüzoğlu, G., 2010. Removal of nitrate from aqueous solutions by
activated carbon prepared from sugar beet bagasse. Bioresour. Technol. 101,
1675–1680.
DeSimone, L.A., Howes, B.L., 1998. Nitrogen transport and transformations in a
shallow aquifer receiving wastewater discharge: a mass balance approach.
Water Resour. Res. 34, 271–285.
Ergas, S.J., Reuss, A.F., 2001. Hydrogenotrophic denitrification of drinking water
using a hollow fibre membrane bioreactor. J. Water Supply: Res. Technol.-AQUA
50, 161–171.
Fernández-Nava, Y., Marañón, E., Soons, J., Castrillón, L., 2010. Denitrification of high
nitrate concentration wastewater using alternative carbon sources. J. Hazard.
Mater. 173, 682–688.
Ghafari, S., Hasan, M., Aroua, M.K., 2008. Bio-electrochemical removal of nitrate
from water and wastewater – a review. Bioresour. Technol. 99, 3965–3974.
Guerin, T.F., Horner, S., McGovern, T., Davey, B., 2002. An application of permeable
reactive barrier technology to petroleum hydrocarbon contaminated
groundwater. Water Res. 36, 15–24.
Haugen, K.S., Semmens, M.J., Novak, P.J., 2002. A novel in situ technology for the
treatment of nitrate contaminated groundwater. Water Res. 36, 3497–3506.
Huang, G., Fallowfield, H., Guan, H., Liu, F., 2012. Remediation of nitrate–nitrogen
contaminated groundwater by a heterotrophic–autotrophic denitrification
(HAD) approach in an aerobic environment. Water Air Soil Pollut. 223, 4029–
4038.
Jha, D., Bose, P., 2005. Use of pyrite for pH control during hydrogenotrophic
denitrification using metallic iron as the ultimate electron donor. Chemosphere
61, 1020–1031.
Lee, K., Rittmann, B.E., 2002. Applying a novel autohydrogenotrophic hollow-fiber
membrane biofilm reactor for denitrification of drinking water. Water Res. 36,
2040–2052.
Levallois, P., Thériault, M., Rouffignat, J., Tessier, S., Landry, R., Ayotte, P., Girard, M.,
Gingras, S., Gauvin, D., Chiasson, C., 1998. Groundwater contamination by
nitrates associated with intensive potato culture in Qúebec. Sci. Total Environ.
217, 91–101.
Liu, H., Jiang, W., Wan, D., Qu, J., 2009. Study of a combined heterotrophic and sulfur
autotrophic denitrification technology for removal of nitrate in water. J. Hazard.
Mater. 169, 23–28.
Liu, S.J., Zhao, Z.Y., Li, J., Wang, J., Qi, Y., 2013. An anaerobic two-layer permeable
reactive biobarrier for the remediation of nitratecontaminated groundwater.
Water Res. 47, 5977–5985.
Liu, F., Huang, G., Fallowfield, H., Guan, H., Zhu, L., Hu, H., 2014. Study on
Heterotrophic–Autotrophic Denitrification Permeable Reactive Barriers (HAD
PRBs) for Groundwater In Situ Remediation. Springer, Germany.
Mendes, M.P., Ribeiro, L., 2010. Nitrate probability mapping in the northern aquifer
alluvial system of the river Tagus (Portugal) using Disjunctive Kriging. Sci. Total
Environ. 408, 1021–1034.
Moon, H.S., Shin, D.Y., Nam, K., Kim, J.Y., 2008. A long-term performance test on an
autotrophic denitrification column for application as a permeable reactive
barrier. Chemosphere 73, 723–728.
Mousavi, S., Ibrahim, S., Aroua, M.K., 2012. Sequential nitrification and
denitrification in a novel palm shell granular activated carbon twin-chamber
upflow bio-electrochemical reactor for treating ammonium-rich wastewater.
Bioresour. Technol. 125, 256–266.
Obiri-Nyarko, F., Grajales-Mesa, S.J., Malina, G., 2014. An overview of permeable
reactive barriers for in situ sustainable groundwater remediation. Chemosphere
111, 243–259.
Pastén-Zapata, E., Ledesma-Ruiz, R., Harter, T., Ramírez, A.I., Mahlknecht, J., 2014.
Assessment of sources and fate of nitrate in shallow groundwater of an
agricultural area by using a multi-tracer approach. Sci. Total Environ. 470–471,
855–864.
Phillips, D.H., 2009. Permeable reactive barriers: a sustainable technology for
cleaning contaminated groundwater in developing countries. Desalination 248,
352–359.
Robertson, W.D., Vogan, J.L., Lombardo, P.S., 2008. Nitrate removal rates in a 15-
year-old permeable reactive barrier treating septic system nitrate. Groundw.
Monit. Remediat. 28, 65–72.
Rust, C.M., Aelion, C.M., Flora, J.R.V., 2002. Laboratory sand column study of
encapsulated buffer release for potential in situ pH control. J. Contam. Hydrol.
54, 81–98.
Sacchi, E., Acutis, M., Bartoli, M., Brenna, S., Delconte, C.A., Laini, A., Pennisi, M.,
2013. Origin and fate of nitrates in groundwater from the central Po plain:
Insights from isotopic investigations. Appl. Geochem. 34, 164–180.
Salvestrin, H., Hagare, P., 2009. Removal of nitrates from groundwater in remote
indigenous settings in arid Central Australia. Desalin. Water Treat. 11, 151–156.
SBTS, 1993. Quality Standard for Ground Water (GB/T 14848-93). State Bureau of
Technical Supervision, Beijing.
Schipper, L.A., Vojvodic-Vukovic, M., 2000. Nitrate removal from groundwater and
denitrification rates in a porous treatment wall amended with sawdust. Ecol.
Eng. 14, 269–278.
Schipper, L.A., Barkle, G.F., Vojvodic-Vukovic, M., 2005. Maximum rates of nitrate
removal in a denitrification wall. J. Environ. Qual. 34, 1270–1276.
Shin, K.H., Cha, D.K., 2008. Microbial reduction of nitrate in the presence of
nanoscale zero-valent iron. Chemosphere 72, 257–262.
Shomar, B., Osenbrück, K., Yahya, A., 2008. Elevated nitrate levels in the
groundwater of the Gaza strip: distribution and sources. Sci. Total Environ.
398, 164–174.
Shrimali, M., Singh, K.P., 2001. New methods of nitrate removal from water.
Environ. Pollut. 112, 351–359.
Soares, M.I.M., Brenner, A., Yevzori, A., Messalem, R., Leroux, Y., Abeliovich, A., 2000.
Denitrification of groundwater: pilot-plant testing of cotton-packed bioreactor
and post-microfiltration. Water Sci. Technol. 42, 353–359.
Su, C., Puls, R.W., 2007. Removal of added nitrate in the single, binary, and ternary
systems of cotton burr compost, zerovalent iron, and sediment: implications for
groundwater nitrate remediation using permeable reactive barriers.
Chemosphere 67, 1653–1662.
Tang, Y., Zhou, C., Ziv-El, M., Rittmann, B.E., 2011. A pH-control model for
heterotrophic and hydrogen-based autotrophic denitrification. Water Res. 45,
232–240.
Till, B.A., Weathers, L.J., Alvarez, P.J.J., 1998. Fe(0)-supported autotrophic
denitrification. Environ. Sci. Technol. 32, 634–639.
van Rijn, J., Tal, Y., Schreier, H.J., 2006. Denitrification in recirculating systems:
theory and applications. Aquacult. Eng. 34, 364–376.
Volokita, M., Abeliovich, A., Soares, M.I.M., 1996a. Denitrification of groundwater
using cotton as energy source. Water Sci. Technol. 34, 379–385.
Volokita, M., Belkin, S., Abeliovich, A., Soares, M.I.M., 1996b. Biological
denitrification of drinking water using newspaper. Water Res. 30, 965–971.
Wang, Q., Feng, C., Zhao, Y., Hao, C., 2009. Denitrification of nitrate contaminated
groundwater with a fiber-based biofilm reactor. Bioresour. Technol. 100, 2223–
2227.
Westerhoff, P., James, J., 2003. Nitrate removal in zero-valent iron packed columns.
Water Res. 37, 1818–1830.
Zhang, Y., Li, Y., Li, J., Sheng, G., Zhang, Y., Zheng, X., 2012. Enhanced Cr(VI) removal
by using the mixture of pillared bentonite and zero-valent iron. Chem. Eng. J.
85–186, 243–249.
Zhang, X., Xu, Z., Sun, X., Dong, W., Ballantine, D., 2013. Nitrate in shallow
groundwater in typical agricultural and forest ecosystems in China, 2004–2010.
J. Environ. Sci. 25, 1007–1014.