Nitrous oxide emissions from wastewater treatment processes

Abstract

Nitrous oxide (N₂O) emissions from wastewater treatment plants vary substantially between plants, ranging from negligible to substantial (a few per cent of the total nitrogen load), probably because of different designs and operational conditions. In general, plants that achieve high levels of nitrogen removal emit less N₂O, indicating that no compromise is required between high water quality and lower N₂O emissions. N₂O emissions primarily occur in aerated zones/compartments/periods owing to active stripping, and ammonia-oxidizing bacteria, rather than heterotrophic denitrifiers, are the main contributors. However, the detailed mechanisms remain to be fully elucidated, despite strong evidence suggesting that both nitrifier denitrification and the chemical breakdown of intermediates of hydroxylamine oxidation are probably involved. With increased understanding of the fundamental reactions responsible for N₂O production in wastewater treatment systems and the conditions that stimulate their occurrence, reduction of N₂O emissions from wastewater treatment systems through improved plant design and operation will be achieved in the near future.

1. Introduction

Nitrous oxide (N₂O) is a potent greenhouse gas, which accounts for 7.9 per cent of the global anthropogenic greenhouse gas emissions in 2004 [1]. It is also predicted to be the most dominant ozone-depleting substance in the twenty-first century [2]. Since 1750, the atmospheric N₂O concentration has increased by about 16 per cent, from around 270 ppb, to 319 ppb in 2005. Human activity has been responsible for 40–50% of the annual increase in N₂O emissions over its pre-industrial levels [1]. While agriculture is the major contributor accounting for 80 per cent of the anthropogenic N₂O source, other contributors include biomass and fossil combustion, manure management, adipic acid and nitric acid production and waste management [1,3].

Since the first published data by Czepiel et al. [4], reporting N₂O emissions from a wastewater treatment plant, awareness and concern of N₂O emissions during wastewater treatment have grown significantly among urban water authorities. Owing to the complexity involved in measuring N₂O emissions from full-scale plants and the lack of standardized measurement methods, N₂O emissions for the wastewater sector have been estimated based on models without the input of measured data. The Environmental Protection Agency of the United States [5] reported that N₂O from the wastewater sector accounts for about 3 per cent of N₂O emissions from all sources and ranks as the sixth largest contributor. Similarly, the Intergovernmental Panel on Climate Change also reports that N₂O emissions from wastewater account for approximately 2.8 per cent of the total anthropogenic sources [1]. Global N₂O emissions from wastewater treatment are expected to increase by approximately 13 per cent between 2005 and 2020.

N₂O is mainly released during biological nitrogen removal in biological nutrient removal (BNR) plants. There are various configurations of BNR plants that can achieve high levels of nitrogen removal from wastewater by promoting nitrification and denitrification in different reaction zones. N₂O is a known obligatory intermediate in the heterotrophic denitrification pathway and is also produced by autotrophic nitrifying bacteria, mainly ammonia-oxidizing bacteria (AOB) [6], as a by-product.

The microbial nitrogen transformation processes in a wastewater treatment plant are fundamentally the same as in other environments such as soil, marine and freshwater habitats. However, unlike most other environments, wastewater treatment plants are engineered systems designed to achieve high nitrogen conversion rates. There are several key features that distinguish these plants from other environments:

• — Domestic wastewater usually contains relatively high concentrations of nitrogen, around 20–70 mg l⁻¹ total nitrogen as N. In order to attain almost complete nitrogen removal within 3–8 h, high nitrogen loading rates are applied, incurring relatively high nitrification and denitrification rates [7]. These are expected to impact on the rate of N₂O production.

• — Bacterial communities in the plants are subjected to rapid changes in process conditions that are applied to promote aerobic or anoxic microbial reactions. Such rapid changes in environmental conditions probably cause physiological stress to both the nitrifying and denitrifying communities, with the potential to induce transient behaviours.

• — Active aeration is used to induce aerobic conditions. The aeration systems are engineered to efficiently provide oxygen to the bioreactor, which also enables efficient transfer of N₂O from the liquid phase to the gas phase. Therefore, any temporary imbalance between N₂O production and consumption could result in accumulation and then stripping of N₂O during aeration.

• — Given that wastewater treatment systems are highly engineered systems, there are opportunities to mitigate N₂O emissions by improving process design and/or operational conditions.

In this paper, we review the key outcomes arising from the research on N₂O production and emissions from wastewater systems. Following a brief description of the design and operation of wastewater treatment systems, the methods for measuring N₂O in wastewater systems and the emission rates thus far measured are summarized. This is followed by discussions on the key metabolic pathways contributing to N₂O production, and the most important influencing factors. Finally possible mitigation strategies are discussed.

2. Design and operation of biological wastewater treatment plants

(a). Activated sludge systems for biological wastewater treatment

Activated sludge is the most widely used process for biological treatment of wastewater. This process uses a microbial community suspended in wastewater to metabolize the biodegradable organic and inorganic components. The microbial community usually clumps together forming three-dimensional aggregates or flocs, known as activated sludge. The sludge and wastewater mixture is called the mixed liquor and the treatment process takes place in a biological reaction tank (bioreactor). At the end of the biological treatment process, the mixed liquor is passed into the clarifier where the sludge is settled and separated from the treated water (figure 1a) [7]. The latter is discharged as the effluent. Most of the settled sludge is returned to the bioreactor, with a hydraulic flow rate that is comparable with that of the influent flow. A small fraction of the sludge, called the waste activated sludge, is removed and disposed of after several steps of sludge treatment (figure 1a). The rate of sludge wastage determines the average amount of time that the sludge is retained in the activated sludge system, and is termed the solids retention time. The key components typically removed during the activated sludge treatment process are solids, organic carbon compounds, nitrogen and phosphorus.

Figure 1.

Diagram of (a) a conventional activated sludge system; (b) a modified Ludzack–Ettinger system; (c) an oxidation ditch; and (d) a sequencing batch reactor system.

(b). Biological processes for nitrogen removal

Nitrogen in wastewater is present in the form of complex organic nitrogen compounds, ammonium (NH₄⁺), and low (often negligible) levels of nitrite (NO₂⁻) and nitrate (NO₃⁻) [7]. The organic nitrogen fraction such as amino acids, amino sugars and proteins are readily converted to NH₄⁺ by microbial degradation in sewer systems and in the bioreactor. In conventional BNR plants, NH₄⁺ is first converted to NO₂⁻ and NO₃⁻ via autotrophic nitrification, which is followed by the reduction of NO₃⁻ and NO₂⁻ by heterotrophic denitrification to form N₂.

The bioreactor used for nitrogen removal provides conditions enabling both nitrification and denitrification. Aerobic conditions are required for nitrification, whereas a sufficient amount of organic carbon compound is required to support denitrification under anoxic conditions. To achieve this, conventional BNR plants are usually configured as continuous systems whereby wastewater flows through the denitrification and nitrification processes, which are separated into either different compartments or zones.

In a typical modified Ludzack–Ettinger configuration, an anoxic compartment/zone precedes the aerobic compartment (figure 1b) [7]. At the end of the aerobic compartment, the nitrified wastewater containing NO₃⁻ is re-circulated back to the anoxic compartment with a flow rate that is a few times that of the wastewater influent. Wastewater is also fed to the anoxic compartment, which provides the organic carbon for denitrification. A wide range of solids retention time (10–30 days) depending on treatment needs can be applied. There are many variants of this configuration. For example, another pair of anoxic and aerobic compartments can be added to the end of the bioreactor shown in figure 1b, to form a four-stage anoxic–aerobic–anoxic–aerobic process. With this design, wastewater is often fed to both the first and second anoxic compartments, thus resulting in a step-feed process. In all cases, the sludge, where micro-organisms reside, is passed between anoxic and aerobic conditions frequently (in hours and in many cases even less than an hour). The dissolved oxygen (DO) concentration in the aerobic compartment is typically within the range of 0.5–2 mg l⁻¹. Although, in some cases could be outside of this range, particularly when the DO is not controlled automatically. In comparison, DO in the anoxic compartment is usually not detectable. However, a limited amount of oxygen is brought into the anoxic compartment through the recirculation stream(s) and through natural surface oxygen transfer.

Figure 1c shows an oxidation ditch system, which is also commonly used for BNR [7]. Oxidation ditches are usually equipped with horizontal brush aerators to provide aeration and also to move the mixed liquor along the ditch at a relatively high velocity (0.25–0.35 m s⁻¹) [8]. Each pass of mixed liquor in the ditch typically lasts for several minutes. A relatively high DO concentration is obtained at or close to the aerator and anoxic conditions develop further away from the aerator. The high recirculation flow and the large tank volume dampen the load variations, giving rise to more stable operating conditions in comparison with the modified Ludzack–Ettinger configuration (figure 1b). A further feature of an oxidation ditch is that the DO is typically low (e.g. around 0.5 mg l⁻¹), favouring simultaneous nitrification and denitrification. Low DO conditions allow a buildup of an oxygen concentration gradient within the microbial flocs as a result of diffusion limitation. Nitrifiers reside at the outer layer of the flocs where there is sufficient oxygen supply, whereas denitrifiers can remain active in the anoxic zone of the flocs allowing nitrification and denitrification to occur simultaneously [9].

Unlike continuous flow systems outlined above, sequencing batch reactors can also be used to achieve the removal of nitrogen and organic carbon. Aerobic and anoxic conditions are separated by time instead of space [10] (figure 1d). All the phases in continuous systems that are spatially separated are provided in a single reactor. A sequencing batch reactor mimics a plug-flow continuous system producing significant concentration gradients of substrates and products with time. This clearly contrasts with the operational conditions found in an oxidation ditch. When a low DO (e.g. 0.5 mg l⁻¹) is provided during an aerobic period of a sequencing batch reactor cycle, simultaneous nitrification and denitrification can also be encouraged.

3. Nitrous oxide measurement in wastewater treatment processes

(a). Gas-phase nitrous oxide measurement

In full-scale wastewater treatment plants, the N₂O emitted from activated sludge tanks is usually captured using a closed floating chamber. This technique was originally adapted from emission measurements of solid surfaces. It was first used to measure N₂O flux from liquid surfaces in a municipal wastewater treatment plant located in Durham, New Hampshire in USA [4]. During aeration, dissolved N₂O was stripped from the liquid phase into the gas; during non-aerated phases, air was blown into the headspace of the chamber for sampling. Owing to the lack of online N₂O measurement at that time, samples were grabbed from the headspace of the chamber into 20 ml nylon syringes at specific time intervals. Analysis for N₂O was accomplished using a gas chromatograph (GC) with an electron capture detector. A similar approach was applied in full-scale studies [11] of an intermittent activated sludge process in Japan. An air pump was used to collect part of the emitted gas from a capture chamber into a gas sampling bag. During the anoxic period, argon was supplied into the chamber as a sweeping gas.

Although the emitted N₂O can be captured through the floating chamber, the off-line sampling (grab samples) do not capture the dynamic changes in the N₂O emission profiles, as will be further discussed. This can result in over- or underestimation of the N₂O emissions. Therefore, online, continuous monitoring of N₂O has been employed in recent years for accurate quantification of N₂O emissions from wastewater treatment systems. The types of online sensors include an infrared analyser [12–15], chemiluminescence [6], a Fourier transform infrared analyser [16] and mass spectrometry [17,18]. Among these, the infrared analyser with a broad N₂O measurement range of up to 2000 ppm is the most commonly used method. However, chemiluminescence has a higher sensitivity with a detection limit at parts per trillion levels.

In addition to temporal variations, spatial variations in N₂O emissions should also be considered, especially for continuous processes (figure 1b,c). Ideally, multiple hoods should be used to measure N₂O emissions from all zones simultaneously. Although not desirable, variations could also be reasonably captured by moving the single hood between zones. For sequencing batch reactor systems (figure 1d), a single location is theoretically adequate, although in practice multiple locations are also preferred to cover possible spatial variation of fluxes.

The N₂O emission factor is typically represented as the ratio between the mass of emitted N₂O-N (kg-N d⁻¹) and the amount of influent total Kjeldahl nitrogen load (kg-N d⁻¹). In some cases, the emission factors are represented as the ratio between the mass of N₂O-N emitted and the amount of N removed through nitrification and denitrification in the treatment plant. The mass of emitted N₂O-N is calculated from the measured N₂O concentration, the gas flow rate out of the chamber and the covered cross-sectional area [19]. For aerated zones, the gas flow out of the chamber is equal to the air flow for aeration and is usually recorded by each plant. For non-aerated zones, the gas flow through the chamber can be recorded with a rotameter.

(b). Liquid-phase nitrous oxide measurement

Measurement of liquid-phase N₂O using off-line grab samples followed by GC analysis has been used in both laboratory scale reactors and full-scale plants [4,15,20–22]. A liquid sample containing N₂O is injected into a vacuum vial and allowed to reach liquid–gas equilibrium. The gas-phase N₂O concentration (C_gas) in the vial is then measured and the liquid-phase N₂O (C_liquid) concentration is calculated based on Henry's law. The total N₂O concentration in the sample is obtained by dividing the total amount of N₂O in both the gas and liquid phases by the total liquid volume.

Continuous monitoring of the dissolved N₂O concentration can be done using N₂O microsensors. Kampschreur et al. [6] used a modified Clark electrode (Unisense, Denmark) to measure the liquid-phase N₂O in two laboratory scale reactors. Foley et al. [23] measured the liquid-phase N₂O in seven full-scale plants in Australia using the same type of microsensor. The Clark-type sensor has an internal reference and a guard cathode. During measurement, N₂O penetrates through the sensor tip membrane and is reduced at the metal cathode surface. The sensor is connected to a high-sensitivity pico-ammeter, which converts the resulting reduction current to a signal. The online signal can be recorded on a laptop. The response of the electrochemical microsensor is known to be linear in the range of 0–1.2 mM [24].

While N₂O microsensors have a low detection limit, the high sensitivity can render it susceptible to interferences especially in full-scale measurements. Combining the analyses of both the microsensor and the GC-vial methods significantly increases the reliability of data.

Similar to the gas-phase analysis, liquid-phase detection at multiple locations is needed to capture the spatial variation in N₂O concentration.

N₂O flux is determined using the liquid-phase measurement [23]. However, this requires the estimation of the mass transfer coefficient between the liquid and gas phases, which is not a straightforward task in full-scale plants [23]. Consequently, the liquid-phase N₂O data are primarily used for understanding N₂O production and emission processes rather than for quantification purposes.

Other parameters such as pH, DO, temperature, total suspended solids and volatile suspended solids (VSS) are often measured at sampling locations and at the wastewater influent for mass balance, correlation analysis of N₂O emission fluxes and for model development.

4. Full-scale emission data

The N₂O emission factor (amount of N₂O-N emitted relative to the nitrogen load) reported thus far for full-scale plants varies substantially, ranging from 0 to 25% (table 1). It should be noted that an emission factor of 1 per cent would already increase the carbon footprint of a wastewater treatment plant by approximately 30 per cent [29]. The large variation in N₂O emissions among the investigated plants was probably owing to the different configurations and operational conditions applied. Additionally, the different monitoring and quantification methods used could have been a contributing factor. The large variation also implies that N₂O emissions from a treatment plant can be reduced through proper plant design and operation. Foley et al. [23] concluded that plants achieving high-level nitrogen removal would emit less N₂O in comparison with nitrifying plants achieving no or low levels of nitrogen removal. This implies that improved water quality and reduced N₂O emissions can be achieved simultaneously.

Table 1.

Nitrous oxide (N₂O) emission factors reported for several full-scale wastewater treatment plants.

type of plant	N2O emission (% of N-influent)	sampling method	remarks	reference
activated sludge plant—primary and secondary treatment (aeration only; 4 ml d−1)	0.035–0.05	weekly grab samples for 15 weeks	N2O was emitted in aerated areas, low N2O flux at non-aerated areas	New Hampshire, USA [4]
activated sludge plant	0.001	grab samples in alternate weeks for 1 year	N2O emissions increased with nitrite and nitrate concentrations	Germany [25]
anoxic–aerobic activated sludge plant (78 Ml d−1)	0.001–0.04	grab samples	N2O emission was dependent on COD:N	Germany [26]
intermittent activated sludge plant (0.2 Ml d−1)	0.01–0.08	collecting gas-phase N2O samples using air bags during four aeration cycles (2 h)	N2O emission decreased with shorter aeration periods	Japan [11]
intermittent activated sludge treatment of municipal sewerage (2.5 and 31 Ml d−1)	0.47 (0.01)	—	—	France [27]
nitritation–anammox sludge digestion liquor treatment	2.3	online measurement during 4 days	N2O emissions increased with decreasing oxygen concentration (aerated stage) and increasing nitrite concentration (anoxic stage)	Netherlands [22]
seven BNR plants	0.6–25 (3.5 + 2.7% average)	grab samples	correlation between N2O emissions and nitrite accumulation was observed	Australia [23]
four treatment plants (completely mixed, plug-flow, membrane bioreactor)	0–0.3	online measurement	NH4-N and DO had impact on N2O emission	France [28]
partial nitritation–anammox sequencing batch reactor (three plants, five reactors)	0.4–0.6	online measurement	N2O emissions were slightly higher than in conventional nitrogen-removal systems	Switzerland [16]
12 BNR plants	0.003–2.59	online measurement	aerobic zones contributed substantially more to N2O fluxes than anoxic zones	USA [14]
four-stage floc-based partial nitritation and anammox process	5.1–6.6	online measurement	high N2O emissions may be partly inherent to a separate nitritation step	Belgium [15]

Many studies show that N₂O is primarily emitted from the aerated zones [14]. Although N₂O is an obligatory intermediate in denitrification, N₂O formed in anoxic zones will largely be dissolved in the liquid phase and this is converted to N₂ through N₂O reduction before it is transferred to the gas phase. In contrast, N₂O formed in aerobic periods is found to be stripped quickly owing to intensive aeration, forming the primary source of N₂O emitted from wastewater treatment systems [14].

5. Nitrous oxide production pathways

N₂O is produced in BNR systems during autotrophic nitrification and heterotrophic denitrification. Although the nitrification step involves both AOB and nitrite-oxidizing bacteria (NOB), it is widely accepted that NOB does not contribute to N₂O production. The key metabolic pathways involved in N₂O production by AOB and denitrifying bacteria in BNR systems are reviewed in this section.

(a). Nitrifier denitrification

Nitrifier denitrification involves the reduction of NO₂⁻ to NO, N₂O and N₂ by autotrophic AOB. However, only genes encoding NO₂⁻ and NO reductase (nirK and nor) are found in the genome of AOB but not N₂O reductase [30–36]. This suggests that N₂O rather than N₂ is the endproduct of the nitrifier denitrification pathway (figure 2). Hydroxylamine (NH₂OH) [38], hydrogen (H₂) [38] and ammonia (NH₃) [39,40] can serve as electron donors for NO₂⁻ and NO reduction by AOB.

Figure 2.

Nitrogen transformation pathways of ammonia-oxidizing bacteria (AOB), nitrite-oxidizing bacteria (NOB) and denitrifying bacteria (modified from Kim et al. [37]). AOB and NOB pathways divided by large dotted line and denitrifying pathway shown in grey.

The nitrifier denitrification pathway plays a key role in N₂O production by AOB, especially under anoxic to suboxic conditions [6,22,41]. Experiments conducted with full-scale sludge show that nitrifier denitrification can contribute up to 83 per cent of the N₂O emissions and this depends on the DO level [42]. Kim et al. [37] also report that the denitrification activity by AOB is the predominant source of N₂O in an activated sludge under nitrifying conditions and they detected simultaneous expression of nirK by AOB.

(b). Autotrophic ammonia oxidation

NH₃ rather than NH₄⁺ is shown to be the true substrate for AOB [43]. Aerobic NH₃ oxidation to NO₂⁻ is a two-step process. NH₃ is first converted to NH₂OH catalysed by a membrane-bound ammonia mono-oxygenase (AMO). This first step requires molecular oxygen and a pair of electrons. The subsequent step is carried out by hydroxylamine oxidoreductase (HAO) in the periplasm to form NO₂⁻, generating two pairs of electrons. One pair is used to support the first step of NH₃ oxidation and the remaining pair is used for energy generation [44].

Extended studies by Igarashi et al. [45], to characterize the structure and function of HAO, suggest that the NH₂OH oxidation is further split into two reactions to allow two electrons to be accepted and transferred simultaneously. The concurrent reaction involves: (i) conversion of NH₂OH to a nitrosyl radical (NOH); and (ii) conversion of NOH to NO₂⁻ [46].

N₂O and NO can be formed from the activity of HAO through the unstable NOH intermediate (figure 2). NO is generated as an intermediate during the enzymatic splitting of NOH to NO₂⁻ [44,47], whereas N₂O is produced through the unstable breakdown of NOH [46].

Despite the fact that this pathway had been postulated for a long time, its relevance to wastewater treatment processes has not been fully confirmed. However, strong evidence demonstrating the potentially significant contribution of this pathway to N₂O production during nitrification is emerging. Increased N₂O production induced by the transition from anoxic to aerobic conditions [48] and high pH [49] are attributed to an increase in NH₃ oxidation rate by AOB. The relationship between the NH₃ oxidation rate and N₂O production rate by AOB was further characterized by Law et al. [50], whereby N₂O production rate was shown to be exponentially correlated to the NH₃ oxidation rate (figure 3). This exponential correlation could be represented by a metabolic model based on N₂O production through the chemical degradation of NOH. This provides evidence that N₂O is produced during increased NH₃ oxidation rates and is most likely produced from the unstable breakdown of NOH during NH₂OH oxidation. This suggestion requires confirmation with further experimental studies.

Figure 3.

Correlation between the specific N₂O production rate and the specific ammonia oxidation rate. Symbols represent experimental data under various conditions. Solid lines are predictions by a model based on the NOH pathway (adapted from Law et al. [50]).

In addition to the chemical breakdown of NOH, biological reduction of NO generated during NH₂OH oxidation could also be a potential source of N₂O. Two molecules of cytochrome c are expressed in AOB for transfer of electrons during NH₂OH oxidation to the electron transport chain [51]. One of the two cytochromes, c₅₅₄, can also act as an NO reductase in vitro [52] and is suggested to produce N₂O from NO generated by the enzyme HAO [53]. NO generated during NH₂OH oxidation could also be reduced by homologue NO reductases (NOR), namely NorS [53]. Indeed, genes encoding NorS are detected in the genome of most AOB [53].

As wastewater treatment systems feature high nitrogen conversion and high nitrogen loading, N₂O production during NH₂OH oxidation either through direct decomposition of NOH or the subsequent reduction of the generated NO could play a crucial role in full-scale systems. In addition, sudden process perturbations leading to transiently increased NH₃ oxidation rates may potentially cause increased N₂O production. However, the relevance of this pathway to the natural environment remains to be verified.

Besides aerobic NH₃ oxidation by AOB, dinitrogen tetroxide (N₂O₄)-dependent NH₃ oxidation is proposed as an alternative pathway for Nitrosomonas to oxidize NH₃ [54]. Catalysed by AMO, NH₃ oxidation to NH₂OH is coupled to N₂O₄ reduction (figure 2) [55,56]. Two moles of NO are formed and released from the cell per mole of N₂O₄ reduced. This enables NH₃ oxidation to proceed under complete anoxic conditions. Similarly, NH₂OH oxidation is also catalysed by HAO, using NO₂⁻ as an electron acceptor to form N₂ as a final product [57]. Under aerobic conditions, molecular oxygen is postulated to have an indirect role by re-oxidizing the NO to form N₂O₄. This NO to N₂O₄/NO₂ conversion is called the NO_x cycle [56]. It is proposed that nitrifier denitrification under oxic conditions plays a role in supplying NO for the NO_x cycle [54].

(c). Heterotrophic denitrification

Heterotrophic denitrification is an enzyme-mediated sequential reduction of NO₃⁻ to N₂ coupled to the oxidation of organic substrates (figure 2). N₂O is an obligate intermediate of heterotrophic denitrification. Under typical denitrifying conditions found in a biological wastewater treatment process, NO and N₂O reductases have higher maximum nitrogen turnover than NO₃⁻ and NO₂⁻ reductases [58]. Wicht [59] estimates that the maximum N₂O reduction rate is almost four times faster than the NO₃⁻ and NO₂⁻ reduction rates. This indicates that N₂O could be completely reduced under anoxic/anaerobic conditions without the occurrence of its accumulation or emission.

However, fluctuations in environmental conditions have been found to lead to inhibition of the N₂O reductase and accumulation of N₂O (see §6). Also, denitrification enzymes are induced during exposure to anaerobic conditions. Under most circumstances, the induction of N₂O reductase appears to lag behind the others resulting in transient accumulation of N₂O [60]. In addition, N₂O has been found to be the principal product for some denitrifiers as there is only approximately a 20 per cent difference in energy loss if denitrification does not proceed to completion [61,62].

The accumulation of N₂O has been found not to result in significant emissions of N₂O because of the lack of active aeration in the anoxic zones. Under such conditions, the air–liquid interface is limited to the surface area of the reactor, which would lead to limited N₂O emission (table 1) given the relatively high solubility of N₂O. However, the accumulated N₂O that is carried over into the aerobic zone will be stripped quickly [14,63]. This emission can be minimized by providing enough anoxic time to allow the temporarily accumulated N₂O to be removed.

6. Key process conditions leading to nitrous oxide emissions

The key process conditions affecting the N₂O production from full-scale wastewater treatment plants are summarized in table 1. These, and a range of other process conditions leading to N₂O emissions are further discussed in this section.

(a). Stripping owing to aeration

In contrast to freshwater, marine or soil environments, N₂O emission from wastewater treatment plants is substantially enhanced owing to the stripping that is induced by active aeration. N₂O is a relatively soluble gas in water with a Henry's law constant of 24 mM atm⁻¹ (at 25°C and 0% salinity) [64] in comparison with 1.3 mM atm⁻¹ (at 25°C and 0% salinity) for oxygen [65]. This implies that N₂O could accumulate to relatively high levels in the liquid phase in the absence of active stripping. For example, Law et al. [49] observed negligible N₂O emission from a nitrifying reactor in non-aerated periods despite its accumulation to 0.5 mg N l⁻¹ in the liquid phase. In contrast, the liquid-phase N₂O was in the range of 0.01–0.03 mg N l⁻¹ in the aerated periods. Here, vigorous aeration employed to promote the activity of nitrifying bacteria resulted in stripping of the dissolved N₂O. Gas-phase N₂O measurements in full-scale plants also show that N₂O emissions are two to three orders of magnitude higher in aerated zones than in non-aerated zones [66]. The emitted N₂O can be either produced under aerobic conditions or accumulated during anoxic conditions preceding the aeration.

(b). Transition between anoxic and aerobic conditions

As described in §2, anoxic and aerobic compartments/periods are engineered in a wastewater treatment system to achieve nitrification and denitrification, respectively. However, a single sludge process consisting of both nitrifiers and denitrifiers is normally employed. The activated sludge is re-circulated between anoxic and aerobic compartments/periods and this would result in exposure of the mixed bacterial community to repeatedly changing conditions. Fluctuations within a compartment can also occur, for example, the DO concentration may decrease owing to increased loading or limitation of the aeration capacity [12,13,67]. Transient changes in DO concentration are shown to cause immediate increase in N₂O production especially from AOB [68–70].

(i). Imposition of anoxia on nitrifying bacteria

It is widely reported that N₂O production from nitrifying cultures is significantly increased during oxygen limitation. Maximum N₂O production rates are observed between DO concentrations of 0.1 and 0.3 mg O₂ l⁻¹ [41,42,71,72]. The response of a nitrifying culture to the transition from aerobic to anoxic conditions was demonstrated by Kampschreur et al. [68]. The N₂O and NO production rates increased instantly upon the imposition of anoxia from fully aerobic condition. The NO production also increased immediately when tested with NO₂⁻ and NH₄⁺ pulsing under both aerobic and anoxic conditions [68]. Nitrifier denitrification by AOB is suggested to be the main pathway contributing to the production of N₂O and NO as both NH₄⁺ and NO₂⁻ were required to be present.

Yu & Chandran [73] further investigated the response of AOB to low DO coupled to NO₂⁻ accumulation at the gene expression and transcription level. During the exponential growth phase of the Nitrosomonas europaea batch culture, mRNA concentrations for ammonia monooxygenase (amoA) and hydroxylamine oxidoreductase (hao) were higher in cultures cultivated at lower DO. In addition, the presence of 280 mg NO₂⁻-N l⁻¹ resulted in elevated concentrations of nirK and norB mRNA for NO₂⁻ reductase and NO reductase, respectively. They postulate that N. europaea increases the efficiency to metabolize NH₃ and NH₂OH under oxygen limitation and also promote the reduction of NO₂⁻ for detoxification purposes when NO₂⁻ accumulates. However, such responses are not observed in stationary phase cells suggesting that the efficiency to metabolize substrate and to detoxify is probably dependent on the physiological growth state of the N. europaea culture.

(ii). Recovery of nitrifying bacteria from anoxic condition

In contrast to the above study, Yu et al. [70] report that it is the recovery from anoxia rather than the transition to anoxia that causes N₂O production from AOB. Such observation was also reported in various full-scale wastewater treatment plants [66]. In an N. europaea pure culture grown in chemostat, NO accumulated under anoxic conditions, however N₂O was produced only during the recovery from anoxic to aerobic conditions [70]. The N₂O production during the transient recovery period correlated positively to the accumulation of NH₄⁺ during anoxia, and the oxygen concentration upon recovery. In addition, the increased N₂O production during the recovery period did not correlate with changes in the gene-expression level. It was therefore concluded that the tendency of nitrifying cultures to produce N₂O is owing to a shift in metabolism from a low specific activity (q < q_max) towards the maximum specific activity (q_max).

Various other studies also report increased N₂O production during increased aeration rate. Sümer et al. [25] found that increased N₂O production coincides with increased oxygen concentration in the activated sludge process. Kampschreur et al. [68] also observed that N₂O production by AOB in a nitritation–anammox process decreased with decreased DO concentrations. However, the mechanisms leading to these observations were not identified.

(iii). Nitrous oxide reduction by denitrifying bacteria during transient aerobic and anaerobic conditions

Similar to nitrifier denitrification, N₂O emission from heterotrophic denitrification is also shown to be the highest under low DO concentrations of around 0.1–0.3 mg O₂ l⁻¹ [18,42,74]. Therefore, transient and dynamic aerobic and anaerobic conditions will likely increase N₂O emission from heterotrophic denitrification. Oxygen inhibits both the synthesis and activity of denitrifying enzymes of Alcaligenes faecalis, in particular the N₂O reductase [18]. The synthesis of the N₂O reductase has a longer lag phase compared with the NO₂⁻ reductase synthesis after the transition from aerobic to anaerobic conditions. In addition, N₂O reductase activity stops immediately during the transition from anaerobic to aerobic conditions, while the activity of NO₂⁻ reductase continues at a lower rate for several hours.

(c). The effect of nitrite, free nitrous acid and pH

(i). Nitrifying bacteria

Hynes & Knowles [75] demonstrate that addition of exogenous NO₂⁻ does not cause an increase in N₂O production from a fully aerobic N. europaea culture. In addition, the optimum pH for the production of NO₂⁻ and N₂O is approximately 8.5, in the investigated pH range of 5.4–9.5, further suggesting that a high free nitrous acid (HNO₂) concentration, the true substrate for NO₂⁻ reduction [76], is not required for higher N₂O production. As the aerobic N₂O production is completely inhibited by acetylene (C₂H₂), the authors suggest that N₂O is predominantly produced through degradation of NOH under aerobic conditions [75]. Increased N₂O production rate of an enriched AOB culture at pH 8.0 when compared with pH 6.0 is also reported by Law et al. [49].

However, contradictory evidence is produced in some recent studies that report elevated N₂O production rates by AOB in the presence of NO₂⁻. Correlation between N₂O production and high NO₂⁻ concentration by AOB is reported in several full-scale studies [15,25,63,67,77]. In laboratory scale studies, NO₂⁻ pulses of 10 mg NO₂⁻-N l⁻¹ are shown to increase N₂O production by a nitrifying mixed culture especially at higher DO concentrations, with eightfold and fourfold increases occurring at DO concentrations of 1.0 and 0.1 mg O₂ l⁻¹, respectively [42]. Kampschreur et al. [68] also reveal that NO₂⁻ pulsing increases N₂O production by an enriched AOB culture under aerobic conditions. The contrasting observations on the effect of NO₂⁻ on N₂O production by AOB are yet to be resolved.

(ii). Denitrifying bacteria

The presence of NO₂⁻ has been shown to affect the activity of N₂O reductase in a denitrifying bacterial culture leading to increased N₂O emission. NO₂⁻ accumulation of up to 10 mg NO₂⁻-N l⁻¹ was identified as a possible cause of N₂O production in denitrifying sludge [78]. However, the effect of NO₂⁻ addition on N₂O accumulation is seen to be highly inconsistent [79]. Schulthess et al. [80] suggest that NO rather than NO₂⁻, which accumulates upon NO₂⁻ addition, is the true inhibitor of N₂O reductase.

Zhou et al. [81] show that HNO₂ rather than NO₂⁻ is responsible for inhibiting the N₂O reductase in an enriched denitrifying biological phosphorus removal system. N₂O reductase activity was inhibited by 50 per cent at a HNO₂ concentration of 0.0007–0.001 mg HNO₂-N l⁻¹ (equivalent to 3–4 mg NO₂⁻ N l⁻¹ at pH 7). However, an internal storage polymer was the sole carbon source available as shown in the study of Zhou et al. [81], which is suggested to be a factor affecting N₂O production (further discussed in §6d(iii)). Since the concentration of NO in the study was not reported, it is unclear whether HNO₂ could have triggered transient NO accumulation to affect the N₂O reductase activity. The high sensitivity of N₂O reductase to low pH (<6.5) [82] also renders it difficult to distinguish the effect of pH and HNO₂ in denitrifying cultures.

(d). Effect of carbon sources

(i). Availability of carbon source

The lack of biodegradable organic carbon is an important factor governing N₂O production during denitrification [83,84]. The availability of organic carbon is typically measured as chemical oxygen demand (COD). For complete denitrification, a COD to N ratio above 4 is required. Under conditions of limited carbon sources, the various denitrification enzymes (NO₃⁻ reductase, NO₂⁻ reductase, NO reductase and N₂O reductase) compete for electrons, potentially resulting in incomplete denitrification.

In an intermittently aerated laboratory scale reactor, approximately 20–30% of influent N was emitted as N₂O when the COD to N ratio was less than 3.5 [78]. Similar observations have also been reported by Kishida et al. [85]. A pure culture study with A. faecalis shows that when carbon sources are limiting, N₂O formation increases by 32–64%, while N₂ production decreases significantly [84]. When excess carbon was supplied to remove electron competition, N₂O formation decreased immediately. On the contrary, it is reported in full-scale studies that only little N₂O generation and emission is observed regardless of carbon deficiency or sufficiency in anoxic zones or aerobic zones [14]. The nitrogen or helium sparging used to induce anoxia in the laboratory scale studies [84,85] may have contributed to the discrepancy between the observation in laboratory scale and full-scale studies. The continuous sparging may have stripped off the dissolved N₂O to render it unavailable for further reduction to N₂. This requires further investigation and verification.

In theory, N₂O and NO are expected to accumulate during COD-limited denitrification as the NO₃⁻ and NO₂⁻ reductases have relatively higher affinity for electrons than the NO and N₂O reductases [86]. However, this may not be generalized for all types of carbon sources as different metabolic pathways are employed for different carbon sources.

(ii). Types of carbon sources

The availability of different types of carbon sources may enrich different groups of bacteria and have different impacts on denitrification efficiency [87,88]. Methanol, ethanol and acetate, and to a lesser extent glycerol or sludge fermentates, are widely used as supplemented carbon sources for enhancing denitrification in BNR plants. While a COD/N of lower than 1.5 resulted in N₂O production in a denitrifying culture fed with acetate and yeast extract [89,90], a COD limitation did not have an apparent impact on ethanol- and methanol-fed denitrifying cultures [19]. In addition, the methanol-fed denitrifiers are shown to have higher susceptibility to oxygen inhibition when compared with the ethanol-fed denitrifiers [19].

On the contrary, pure culture studies with A. faecalis indicate that N₂O production is independent of the energetics of the substrate or the turnover rates of the enzymes. The type of supplemented carbon source (acetate versus butyrate) and the growth rate of the bacteria do not have any impact on overall N₂O production [84]. The discrepancy between different studies may be attributed to the enrichment of different denitrifying populations. Therefore, the types of carbon sources used would affect the types of denitrifiers enriched which potentially have different susceptibility to other operational variables (e.g. NO₂⁻ and O₂ inhibition).

(iii). Consumption of internal storage compound

Systems operated to achieve simultaneous nitrification, denitrification and phosphorus removal can encourage the growth of denitrifiers, such as polyphosphate-accumulating organisms and glycogen-accumulating organisms that are capable of storing organic carbon in the form of polyhydroxybutyrate (PHB). Laboratory scale studies on such systems show that denitrification by glycogen-accumulating organisms leads to increased N₂O emission [17,91,92]. During anaerobic periods, these micro-organisms take up organic carbon for storage and subsequently degrade the PHB stored during aerobic/anoxic periods. Since PHB consumption is the rate-limiting step in these organisms [93], high N₂O emission is possible by organisms growing on storage compounds owing to a slow supply of electrons, resulting in competition for electrons between denitrifying enzymes. Schalk-Otte et al. [84] observed that N₂O accumulation coincides with the onset of storage compound usage upon COD depletion.

On the contrary, in a PHB-degrading denitrifying pure culture, no accumulation of N₂O or nitrite was detected when PHB was used as the sole carbon source [94]. Further confirmation on the relationship between internal storage compounds and N₂O production is essential as the dynamic conditions employed in treatment plants, such as in P-removal processes and bioselectors, are operated to select for organisms that are able to store carbon sources.

(e). Availability of copper ions

Copper is essential for the biosynthesis of N₂O reductase and its availability affects N₂O production in soil and marine environments [62]. Deficiency in copper supply is found to shift the endproduct of heterotrophic denitrification from N₂ to N₂O, whereas replenishing the copper supply reduces N₂O production and increases N₂ production [95,96]. Although copper is demonstrated to increase the N₂O reductase activity and to reduce N₂O production in an activated sludge [97], the availability of copper in wastewater systems and its subsequent effect on N₂O production has thus far not been investigated.

7. Possible mitigation strategies

Although the exact triggers for N₂O production by nitrifying and denitrifying sludge are yet to be fully revealed, and the predominant pathway relating to N₂O production by AOB remains to be elucidated, it is generally observed that sudden process perturbations such as rapid shifts in reactor pH, DO and NH₄⁺ or NO₂⁻ spikes lead to immediate increases in N₂O emissions [68,90,98]. In fact, N₂O emission has been recommended to be used as an indication of biological nitrification failure owing to toxic shock loads or insufficient aeration [13].

It is postulated that full-scale plants that are designed and configured to operate under more stable process conditions, such as oxidation ditches with uniform DO concentrations, produce less N₂O when compared with those that are subject to frequent transitions (such as a modified Ludzack–Ettinger plant; figure 1) [70]. Full-scale studies also report that treatment plants designed and operated to achieve low total nitrogen in their effluents are equipped with design features that result in relatively low N₂O emission levels [67]. These design features include influent flow balancing, high recycle rates, large bioreactor volumes and long solids’ retention time. Large bioreactor volumes and influent flow-balancing facilities equip the system with the ability to buffer loadings and reduce the risk of transient oxygen depletion. High recycle rates also tend to dilute the concentrations of NH₄⁺ and nitrogen intermediates which lessens the effects of nitrification and denitrification, preventing the buildup of NO₂⁻ and NH₄⁺ to levels that may increase N₂O production [67,99,100].

Several mitigation strategies have been trialled in laboratory scale studies to minimize N₂O emissions. Yang et al. [21] demonstrate that NH₄⁺ and NO₂⁻ concentrations in the reactor can be maintained at low concentrations through step feeding, resulting in a 50 per cent reduction in N₂O production. Avoiding transient pH changes under aerobic conditions by slow feeding rather than pulse feeding is shown to significantly reduce N₂O production by an enriched AOB culture [49]. Applying longer solids retention time to increase the AOB biomass concentration (greater than 5 days) and higher DO (>0.5 mg O₂ l⁻¹) is also proposed to minimize N₂O production from nitrification [71]. Pellicer-Nàcher et al. [101] demonstrate the possibility of minimizing N₂O emission through sequential aeration in a membrane-aerated biofilm reactor. Here, the N₂O produced by AOB within the membrane bundle is consumed by heterotrophic bacteria outside the bundle. To minimize N₂O production during denitrification, methanol addition prevented N₂O accumulation by eliminating electron competition from other denitrifying enzymes [90].

These mitigation strategies have so far only been demonstrated in laboratory scale systems. Their effectiveness is yet to be verified through full-scale trials. The research community is making steady progress in gaining understanding of the mechanisms involved in N₂O emission in wastewater treatment systems, which will enable the development of effective mitigation strategies.

8. Conclusions

Despite their relatively small contribution to the overall global greenhouse gas emissions, N₂O emissions from BNR wastewater treatment plants can be very significant in terms of their contribution to the overall carbon footprint of wastewater treatment systems, and should be understood, accounted for and mitigated.

N₂O emissions from wastewater treatment processes vary substantially between plants depending on the design and operation of the plants, and the flow and characteristics of wastewater. These variations indicate that N₂O may be mitigated through proper process design and operation. Indeed, preliminary strategies have been developed but remain to be verified through full-scale applications.

In contrast with many other systems (e.g. soil), where denitrification is revealed to be the primary source of N₂O, autotrophic NH₃ oxidation is found to make relatively more contributions than heterotrophic denitrification in most wastewater treatment plants. This is probably related to the fact that AOB produce N₂O under aerated conditions; and most of the N₂O produced is stripped instantly by aeration. In contrast, denitrifiers produce N₂O primarily under non-aerated conditions. N₂O can remain dissolved in the absence of stripping, giving time for its subsequent reduction to N₂. However, N₂O carried over from non-aerated zones/periods to the aerated zones/periods will probably be stripped there.

The detailed mechanisms involved in N₂O production by AOB remain to be fully elucidated. Both nitrifier denitrification and the breakdown or degradation of nitrification intermediates probably contribute to the overall N₂O production. However, the level of contribution by each of these two processes is unclear and contradictory evidence has been produced. Indeed, their relative contributions could be dependent on process conditions.

Various factors have been reported in the wastewater literature to induce N₂O emissions by AOB and denitrifiers. A detailed understanding of the factors is currently missing.

Future research in the field will focus on both the quantification and reduction of N₂O emissions from various full-scale wastewater treatment plants. Additionally, future studies will reveal the fundamental processes involved in N₂O production by both nitrification and denitrification.