Long-term continuous treatments of synthetic groundwater containing 15.06 ± 0.55 mg-N/L nitrate were conducted using an in situ denitrification and oxidation process under different operating conditions. In the experiments, electrolytic hydrogen and oxygen were injected upstream and downstream in a laboratory-scale aquifer, respectively, and measurements were made for nitrate, nitrite, pH, dissolved oxygen, dissolved hydrogen, total organic carbon (TOC), turbidity, and chromaticity. Experimental results demonstrated that steady-state denitrification and oxygenation of groundwater were achieved over 820 days. During the initial phase of the experiments, nitrite was accumulated at a level of several mg-N/L in a hydrogen-injected zone, but it was almost completely oxidized to nitrate in a downstream oxygen-injected zone. In subsequent experiments, nitrite accumulation was not observed in effluent, and satisfactory treatments were achieved. Effluent concentrations of total nitrate and nitrite, TOC, turbidity, and chromaticity were generally below World Health Organization guidelines for drinking water (11.3 mg-N/L). It should be mentioned that water quality parameters such as TOC, turbidity, and chromaticity were lower in effluent than influent, and no clogging problem was observed. From these results, we concluded that the present in situ process has superior properties in terms of long-term performance and stability.

INTRODUCTION

Nitrate contamination in groundwater, mainly in shallow aquifers, has become a serious global environmental issue. This contamination is mainly due to excessive applications of synthetic nitrogen fertilizers and/or inappropriate managements of livestock waste (Keeney 1986; Follett 1989). To treat nitrate-contaminated groundwater, different biological denitrification processes have been proposed (Chang et al. 1999; Nuhoglu et al. 2002; Ergas & Rheinheimer 2004). In the processes, nitrate is biologically reduced to a non-toxic product, namely nitrogen gas, through the addition of organic or inorganic electron donors (Ghafari et al. 2008; Sahu et al. 2009; Karanasios et al. 2010).

Among biological processes, in situ denitrification using hydrogen gas as an electron donor is considered one of the most promising and cost-effective alternatives (Haugen et al. 2002; Lee & Rittmann 2002), because the secondary pollution by electron donors such as methanol (Du Toit & Davies 1973), ethanol (Janda et al. 1988), and sucrose (Mercado et al. 1988) is not anticipated. Moreover, it is thought that clogging problems in aquifers may be avoided due to smaller growth rates of autotrophs in comparison with heterotrophs (Komori & Sakakibara 2008; Komori & Sakakibara 2009).

In our prior studies (Ye et al. 2013), the effectiveness of an in situ process with the injection of hydrogen and oxygen gases was demonstrated. In this process, nitrate was reduced to nitrogen gas with slight accumulations of nitrite in a hydrogen-injected zone, and oxygen was supplied subsequently in a downstream oxygen-injected zone with the complete oxidation of nitrite to nitrate as well as reductions of total organic carbon (TOC). However, long-term performances and stability were not sufficiently investigated. In this study, continuous experiments using a laboratory-scale aquifer were conducted over 800 days under different nitrate loading conditions, and measurements were made for water quality parameters such as nitrate, nitrite, pH, TOC, turbidity, and chromaticity to evaluate long-term performance and stability.

MATERIALS AND METHODS

Experimental apparatus setup

Figure 1 shows the experimental setup of the laboratory-scale aquifer used for the continuous removal of nitrate and nitrite. The dimensions were 1950 mm in length, 25 mm in width, and 780 mm in height. The aquifer was filled with glass beads (φ = 2 mm) as the carrier of micro-organisms. The total and effective liquid volumes of the aquifer were about 40 L and 16 L, respectively. The hydraulic retention time (HRT) was calculated based on the effective liquid volume. A total of 18 sampling ports were allocated in horizontal and vertical directions within 300 mm intervals (shown in Figure 1).

Figure 1

Schematic diagram of experimental apparatus.

Figure 1

Schematic diagram of experimental apparatus.

To achieve a uniform flow, synthetic groundwater was fed and withdrawn continuously using peristaltic pumps. The flow rate of each peristaltic pump was controlled almost every day. However, as shown in Figure 1, the pumps were not set for two withdrawal lines to keep the groundwater flow rate of the influent and effluent at a constant value. Additionally, the effluent from the aquifer was discharged through an overflow weir to maintain the water level.

In these experiments, hydrogen and oxygen gases were produced using a solid polymer electrolyte membrane electrode (Komori & Sakakibara 2008), and they were then injected at the bottom of ports 6 and 15 using peristaltic pumps. To avoid algal growth, the aquifer and every tube were covered with a light-shielding curtain or aluminium foil. The apparatus was placed in a chamber in which the temperature was maintained around 20 °C.

Continuous experiments

Synthetic groundwater was prepared by dissolving sodium nitrate at 15.06 ± 0.55 mg-N/L and trace nutrients (Chang et al. 1999; Sahu et al. 2009; Ye et al. 2013) with inorganic carbon (IC) as NaHCO3 or CO2 in deionized water. All experiments were conducted until steady-stable conditions were achieved. Conditions of the continuous experiments (including our former studies Runs 1–3) are given in Table 1. The composition of the synthetic groundwater is given in Table 2. Prior to the experiments, 20 g of surface soil was obtained from the university's campus, and it was put in 1000 mL of deionized water. A supernatant of this soil/water mixture (500 mL) and an enriched denitrifying culture (500 mL) (Komori & Sakakibara 2009) were inoculated into the aquifer. After that, batch treatments of synthetic groundwater were conducted for 18 days in the same manner for gas injections in continuous experiments.

Table 1

Experimental conditions

Run no.1234567
Flow rate (Q) (L/day) 5.8 ± 0.3 11.3 ± 0.4 
HRT (day) 2.7 ± 0.2 1.4 ± 0.1 
Temperature (°C) 20 ± 2 
IC (as C, mg/L) 10 ± 1 101 ± 6 250 ± 100 125 ± 10 
H2 gas (mL/day) 360 ± 20 580 ± 100 930 ± 85 
O2 gas (mL/day) 220 ± 60 430 ± 30 
Run no. 8 9 10 11 12 13  
Flow rate (Q) (L/day) 22.5 ± 1.1 34.3 ± 2.1 44.8 ± 0.7 35.9 ± 1.4 23.2 ± 1.2 11.4 ± 0.3  
HRT (days) 0.71 ± 0.04 0.47 ± 0.03 0.36 ± 0.01 0.44 ± 0.02 0.69 ± 0.04 1.4 ± 0.1  
Temperature (°C) 20 ± 2  
IC (as C, mg/L) 125 ± 10  
H2 gas (mL/day) 1550 ± 80 1990 ± 80 2150 ± 30 2220 ± 60 1560 ± 40 930 ± 50  
O2 gas (mL/day) 860 ± 20 940 ± 110 1050 ± 65 1130 ± 40 790 ± 35 465 ± 60  
Run no.1234567
Flow rate (Q) (L/day) 5.8 ± 0.3 11.3 ± 0.4 
HRT (day) 2.7 ± 0.2 1.4 ± 0.1 
Temperature (°C) 20 ± 2 
IC (as C, mg/L) 10 ± 1 101 ± 6 250 ± 100 125 ± 10 
H2 gas (mL/day) 360 ± 20 580 ± 100 930 ± 85 
O2 gas (mL/day) 220 ± 60 430 ± 30 
Run no. 8 9 10 11 12 13  
Flow rate (Q) (L/day) 22.5 ± 1.1 34.3 ± 2.1 44.8 ± 0.7 35.9 ± 1.4 23.2 ± 1.2 11.4 ± 0.3  
HRT (days) 0.71 ± 0.04 0.47 ± 0.03 0.36 ± 0.01 0.44 ± 0.02 0.69 ± 0.04 1.4 ± 0.1  
Temperature (°C) 20 ± 2  
IC (as C, mg/L) 125 ± 10  
H2 gas (mL/day) 1550 ± 80 1990 ± 80 2150 ± 30 2220 ± 60 1560 ± 40 930 ± 50  
O2 gas (mL/day) 860 ± 20 940 ± 110 1050 ± 65 1130 ± 40 790 ± 35 465 ± 60  
Table 2

Composition of synthetic groundwater

ChemicalConcentration (mg/L)
NaNO3 15.06 
IC 10–250 
K2HPO4 1.76 
KH2PO4 2.08 
MgSO4·7H24.00 
NaCl 0.96 
CaCl2 1.12 
FeCl3·6H21.92 
ChemicalConcentration (mg/L)
NaNO3 15.06 
IC 10–250 
K2HPO4 1.76 
KH2PO4 2.08 
MgSO4·7H24.00 
NaCl 0.96 
CaCl2 1.12 
FeCl3·6H21.92 

As given in Table 1, 13 different runs (Runs 1–13) were conducted with a constant temperature and influent nitrate concentration, while other parameters such as the flow rate of liquid and gases and the IC were changed. The concentration of NaHCO3/CO2 (i.e., 101 mg-C/L) was set up referring to a concentration range of 2.81–484.8 mg-C/L in groundwater, as reported in prior studies (Herczeg et al. 1991a; Herczeg et al. 1991b; Cai et al. 2003).

In continuous treatments, the mass balance of hydrogen was evaluated in zone ‘ANAERO’ near the injection port of hydrogen, as shown on the left side of Figure 1.

Sampling and analytical methods

Liquid samples were taken from the influent, ports 1–18, and the effluent, while gas samples were collected from gas influent injection points and the gas outlet point. Liquid samples for the ion chromatography analysis were filtered using a 0.45-μm syringe-driven membrane filter 25CS045AN (Advantec), while no filtration was performed for other samples.

Nitrate , nitrite , chloride ion (Cl), and sulfate were measured by ion chromatography HIC-10A (Shimadzu). Dissolved hydrogen and a potential intermediate product – nitrous oxide gas – were measured by gas chromatography GC-8A (Shimadzu). TOC, dissolved oxygen (DO), and pH were measured by TOC-5000A (Shimadzu), an Ultra DO meter (Central Kagaku Corp.), and pH510 (Eutech Instruments), respectively. Turbidity and chromaticity were measured by a digital turbidity/chromaticity meter WA-PT-4DG (Kyoritsu Chemical Check Lab. Corp.). The temperature in the thermostatic chamber was monitored using a temperature recorder TR-71D (T&D Corp.).

Mass balance of hydrogen

In this study, the mass balance of hydrogen gas was evaluated according to the concentration at zone ‘ANAERO’, which is shown in Figure 1. The overall denitrification and oxygen uptakes were represented by the following reactions (Eggers & Terlouw 1979; Kurt et al. 1987; Szekeres et al. 2001; Ye et al. 2013): 
formula
1
 
formula
2
The mass balance equation for hydrogen gas is as follows: 
formula
3
where QH2 is the amount of hydrogen gas injected per day (mmol/day), QDe and QO2 are equivalent amounts of hydrogen used for denitrification and oxygen utilization (mmol/day), and QEx and QEx.gas are those discharged with bulk liquid and gas effluent (mmol/day), respectively. QEx and QEx.gas were measured directly, and the amount of QDe and was calculated by the following equations: 
formula
4
 
formula
5
where Q is the liquid flow rate (L/day), and CNin/CNout and DOin/DOout are the mean nitrate concentration (mg-N/L) and DO (mg/L) of the influent and effluent for the zone ‘ANAERO’, as shown in Figure 1.

RESULTS AND DISCUSSION

Removal of nitrate and nitrite

Figure 2 illustrates temporal changes of nitrate and nitrite against flow rates, as well as the injection rates of hydrogen and oxygen. In Run 1, nitrate concentrations decreased 0.75 m from the inlet, and they increased with the covered distance downstream. Concentrations of nitrite increased 0.75 m from the inlet and decreased thereafter. This means that incomplete denitrification with nitrite accumulation occurred in a hydrogen-injected zone, but nitrite was almost oxidized to nitrate in the following oxygen-injected zone. The accumulation of nitrite was due to a relatively low buffer, as shown in Table 1. Since nitrite is much more toxic than nitrate (Comly 1945; Amenu & Kumar 2008), oxidation of nitrite to nitrate is advantageous for the aim of reducing the toxicity of the effluent.

Figure 2

Temporal changes of liquid and gas flow rate and denitrification performance.

Figure 2

Temporal changes of liquid and gas flow rate and denitrification performance.

In Runs 2–13, the concentration of IC was increased from 10.2 to 250 and then decreased to 125 mg-C/L. The accumulation of nitrite was rarely observed, and very stable denitrification could be achieved. In Run 2, the accumulation of nitrite shrank, and more stable denitrification was observed. From Runs 3 to 6, the injection rate of hydrogen gas was increased while the liquid flow rate was kept stable. In Run 3, around Day 280, effluent nitrate concentrations were decreased further and then stabilized. This may be due to the effectiveness of the in situ denitrification process.

In Run 4, the injection rate of hydrogen gas was modified in a stepwise manner (as indicated by a circle in Figure 4) in the middle of the run. Nitrate concentrations tended to increase, but after the hydrogen injection rate was adjusted to the original state, within 1 week the nitrate concentrations returned to the same level as could be observed before the changes. Therefore, the stability of this process could be proven. In Run 6, where the injection rates of hydrogen gas were sometimes lower than a designed injection rate, some nitrate concentrations in the aquifer increased, but the effluent nitrate was kept at the same level. From Runs 7 to 13 (to verify the performance of the system at different hydraulic load), the liquid flow rate was increased to about two, four, six, and eight times that of the previous runs and then dropped back to two times the original flow rate again. Given the limited gas-holding capacity of the glass beads used in the aquifer, the injection rates of hydrogen and oxygen gases were increased, but by less than about eight times those of the previous runs. Generally, the concentration of nitrate and nitrite in these runs stabilized soon after conditions changed. An exception could be observed for Run 7, which could be due to a slight fluctuation of hydrogen gas injection.

Figure 3 depicts a comparison of the average nitrogen removal and liquid flow rate of each run. In Runs 7–13, the relationship between liquid flow and nitrate in the effluent is shown. Generally, the nitrogen in the effluent exhibited a direct correlation with the increase and decrease of the liquid flow rate. Moreover, the nitrogen in the effluent of the right side bounded by Run 10 was slightly higher than that of the left part. It should be considered that the liquid flow rate of the right part was slightly larger than that of the former part in Figure 3. Furthermore, the nitrogen removal showed an inverse relation to the liquid flow rate. In summary, the removal rate decreased with an increase in the flow rate, but the stability of the laboratory-scale aquifer with varying hydraulic load has been proven. This means that the present process can be applied to a range of different groundwater velocities.

Figure 3

Comparison of nitrogen removal and liquid flow rate of each run.

Figure 3

Comparison of nitrogen removal and liquid flow rate of each run.

Effluent water qualities

Figure 4 illustrates the temporal changes of quality parameters such as DO, turbidity, chromaticity, and TOC. In every experiment, the influent synthetic groundwater contained 7.63 ± 1.01 mg/L DO, which decreased to around 2.0–3.0 mg/L at 0.75 and 1.05 m from the inlet of the aquifer. The effluent contained >15 mg/L due to the oxygen injected under port 15, and full oxygenation was achieved. In conclusion, the DO was consumed and oxygenated by injected hydrogen in the ‘ANAERO’ zone and injected oxygen in the ‘AERO’ zone, respectively. Moreover, DO at 1.05 m from the inlet is slightly higher than at 0.75 m. It may be that part of the injected oxygen was transported upstream by diffusion in the aquifer.

Figure 4

Temporal changes of DO, turbidity, chromaticity, and TOC.

Figure 4

Temporal changes of DO, turbidity, chromaticity, and TOC.

Although the turbidity and chromaticity fluctuated in the influent, they decreased with the distance covered downstream and reached 0.2 NTU (nephelometric turbidity units) and 1.5 PCU (platinum–cobalt colour units), respectively. These values meet the current drinking water standard (World Health Organization 2011). Since glass beads have poor adsorption capacity for every constituent in groundwater (Table 2), it was thought that suspended solids as well as soluble and colloidal constituents exerting chromaticity were adsorbed by, attached to, or decomposed by biofilms that formed on the glass beads (Wik 1999).

The influent solution used in this study contained about 2.4 mg/L TOC on average. Although data variations were observed, the TOC difference between the influent and the effluent was about −0.65 mg/L on average, indicating that TOC contained in the influent or produced in the aquifer was removed during the treatment. As Schnobrich et al. (2007) reported that there was a slight accumulation of TOC (i.e., 0.387 ± 0.25 mg/L as C) in an in situ hydrogenotrophic denitrification process, thus it could be considered that TOC produced in zone ‘ANAERO’ was removed aerobically in the oxygen injection zone (Schnobrich et al. 2007).

Figure 5 shows steady-state profiles of pH, DO, and the total nitrogen concentration of nitrate and nitrite in Runs 1–13. In Run 1, a significant variation of pH was observed due to a small amount of IC (around 10 mg-C/L). In Run 5, the pH value was significantly lower than that of Runs 6–13, which could be due to the excessive injection of CO2 into the influent. For Runs 6–13, pH variations were reduced due to the relatively higher IC concentrations (or buffering capacity) around 125 mg-C/L. Acidic groundwater was fed in Runs 6–13, and pH values increased from about 6.6 to around 7. These results indicate that neutralization of groundwater could be achieved.

Figure 5

Profiles of pH, DO, and nitrogen.

Figure 5

Profiles of pH, DO, and nitrogen.

In Runs 1, 2, and 10, effluent concentrations of total nitrogen were higher than 10 mg-N/L. This is attributable to an insufficient amount of hydrogen that was injected to achieve Reactions (1) and (2), and the high liquid flow rate (which led to low HRT in the aquifer), respectively. In other runs, effluent concentrations of nitrate, nitrite, TOC, turbidity, and chromaticity met WHO guidelines for drinking water quality (World Health Organization 2011).

During the continuous experiments over 2 years, no clogging problem was observed. In addition, a visible thin film was observed on the surface of the glass beads. This was thought to be attributed to small net growth rates of autotrophic microorganisms in the aquifer. Based on the experimental results shown in Figures 25, it can be concluded that the present in situ process has superior performances in terms of stability, effluent water quality, and simplicity of operation. A further kinetic study will be needed to analyse and evaluate the performance more precisely under different operating and design conditions.

Mass balance of hydrogen gas

Nitrous oxide (N2O) is considered an intermediate product during the denitrification process (Knowles 1982). To confirm the complete reduction of nitrate to nitrogen gas, measurements of dissolved nitrous oxide in addition to nitrite have been performed. However, this compound was below the detection limit (0.03%), indicating negligibly small amounts of intermediates.

Figure 6 depicts the results of the mass balance of hydrogen gas in Runs 2–13, based on Equations (3)–(5). In Figure 6, the percentages of the terms on the right side of Equation (3) are shown. Part of the injected hydrogen gas was consumed by denitrification and oxygen utilization, and the rest was discharged to bulk liquid and gas effluent. Based on Figure 6, it can be concluded that a satisfactory mass balance was obtained, and most data indicated that >90% of the injected hydrogen gas was consumed by denitrification (QDe) and DO consumption (QO2) in Runs 2 and 3. In contrast, in Runs 4–13, total fractions considered in Equation (3) were calculated as roughly 80%. According to the observation results of hydrogen injection, hydrogen gas reached the head space of the apparatus. It was thought that this part of the hydrogen gas managed to escape from the aquifer.

Figure 6

Comparison of mass balance of hydrogen gases, where 100% indicates the amount of hydrogen injected.

Figure 6

Comparison of mass balance of hydrogen gases, where 100% indicates the amount of hydrogen injected.

CONCLUSION

In this work, long-term continuous experiments were conducted in order to evaluate the denitrification performance and stability under different treatment conditions. Based on experimental and analytic results, the following conclusions were made:

  1. The laboratory-scale aquifer has excellent adaptability to hydraulic load variation even when the groundwater velocity reaches 1.8 m/day, which is relatively high in an actual situation.

  2. Neutralization of groundwater is possible if the pH of nitrate contaminated groundwater is weakly acidic.

  3. A long-term stable denitrification and oxygenation process was achieved, and most of the hydrogen gas was consumed by denitrification/oxygen utilization.

  4. Observed results demonstrated that, on average, water quality parameters such as TOC, turbidity, and chromaticity get lower in the effluent than the influent. Furthermore, no clogging occurred during the long-term experiment.

These conclusions indicate that the present in situ process has an excellent ability to adapt to changes in hydraulic load, and it shows outstanding attributes for the remediation of weak acidic nitrate-contaminated groundwater. It was demonstrated that the present in situ denitrification and oxygenation process showed superior efficiency in terms of long-term performance and stability. Further study of kinetic analysis will be needed to evaluate the optimum design and operation conditions.

REFERENCES

REFERENCES
Amenu
G. G.
Kumar
P.
2008
A model for hydraulic redistribution incorporating coupled soil-root moisture transport
.
Hydrology and Earth System Sciences
12
(
1
),
55
74
.
Comly
H. H.
1945
Cyanosis in infants caused by nitrates in well water
.
Journal of the American Medical Association
129
(
2
),
112
116
.
Du Toit
P. J.
Davies
T. R.
1973
Dénitrification. Studies with laboratory-scale continuous-flow units
.
Water Research
7
(
3
),
489
500
.
Ergas
S. J.
Rheinheimer
D. E.
2004
Drinking water denitrification using a membrane bioreactor
.
Water Research
38
(
14–15
),
3225
3232
.
Follett
R. F.
1989
Nitrogen Management and Ground Water Protection
.
Elsevier Science
,
New York
,
USA
.
Ghafari
S.
Hasan
M.
Aroua
M. K.
2008
Bio-electrochemical removal of nitrate from water and wastewater – a review
.
Bioresource Technology
99
(
10
),
3965
3974
.
Haugen
K. S.
Semmens
M. J.
Novak
P. J.
2002
A novel in situ technology for the treatment of nitrate contaminated groundwater
.
Water Research
36
(
14
),
3497
3506
.
Herczeg
A. L.
Torgersen
T.
Chivas
A. R.
Habermehl
M. A.
1991b
Geochemistry of ground waters from the Great Artesian Basin, Australia
.
Journal of Hydrology
126
(
3–4
),
225
245
.
Janda
V.
Rudovsky
J.
Wanner
J.
Marha
K.
1988
In situ denitrification of drinking water
.
Water Science and Technology
20
(
3
),
215
219
.
Karanasios
K. A.
Vasiliadou
I. A.
Pavlou
S.
Vayenas
D. V.
2010
Hydrogenotrophic denitrification of potable water: a review
.
Journal of Hazardous Materials
180
(
1–3
),
20
37
.
Keeney
D.
1986
Sources of nitrate to groundwater
.
Critical Reviews in Environmental Control
16
(
3
),
257
304
.
Knowles
R.
1982
Denitrification
.
Microbiological Reviews
46
(
1
),
43
70
.
Komori
M.
Sakakibara
Y.
2009
High-rate hydrogenotrophic denitrification of synthetic groundwater by a fixed-bed bioreactor equipped with spe electrolytic cell
.
Journal of Environmental Engineering (Jsce) G
65
(
3
),
153
163
.
Mercado
A.
Libhaber
M.
Soares
M. I. M.
1988
In situ biological groundwater denitrification: concepts and preliminary field tests
.
Water Science and Technology
20
(
3
),
197
209
.
Nuhoglu
A.
Pekdemir
T.
Yildiz
E.
Keskinler
B.
Akay
G.
2002
Drinking water denitrification by a membrane bio-reactor
.
Water Research
36
(
5
),
1155
1166
.
Sahu
A. K.
Sengupta
S.
Ergas
S. J.
2009
Onsite wastewater denitrification using a hydrogenotrophic hollow-fiber membrane bioreactor
.
Water Environment Research
81
(
7
),
680
686
.
Szekeres
S.
Kiss
I.
Bejerano
T. T.
Soares
M. I. M.
2001
Hydrogen-dependent denitrification in a two-reactor bio-electrochemical system
.
Water Research
35
(
3
),
715
719
.
World Health Organization
2011
Guidelines for Drinking-Water Quality. World Health Organization, Geneva
.
Ye
J.
Nidegawa
Y.
Sakakibara
Y.
2013
An in-situ denitrification and oxidation process with injection of electrolytic hydrogen and oxygen
.
Journal of Japan Society of Civil Engineers, Ser. G (Environmental Research)
69
(
7
),
III_99
III_104
.