OUP user menu

CANON and Anammox in a gas-lift reactor

A. Olav Sliekers, K.A. Third, W. Abma, J.G. Kuenen, M.S.M. Jetten
DOI: http://dx.doi.org/10.1016/S0378-1097(02)01177-1 339-344 First published online: 1 January 2003


Anoxic ammonium oxidation (Anammox) and Completely Autotrophic Nitrogen removal Over Nitrite (CANON) are new and promising microbial processes to remove ammonia from wastewaters characterized by a low content of organic materials. These two processes were investigated on their feasibility and performance in a gas-lift reactor. The Anammox as well as the CANON process could be maintained easily in a gas-lift reactor, and very high N-conversion rates were achieved. An N-removal rate of 8.9 kg N (m3 reactor)−1 day−1 was achieved for the Anammox process in a gas-lift reactor. N-removal rates of up to 1.5 kg N (m3 reactor)−1 day−1 were achieved when the CANON process was operated. This removal rate was 20 times higher compared to the removal rates achieved in the laboratory previously. Fluorescence in situ hybridization showed that the biomass consisted of bacteria reacting to NEU, a 16S rRNA targeted probe specific for halotolerant and halophilic Nitrosomonads, and of bacteria reacting to Amx820, specific for planctomycetes capable of Anammox.

  • Anammox
  • Nitrification
  • Gas-lift reactor
  • Ammonia

1 Introduction

Anoxic ammonium oxidation (Anammox) is an anoxic microbiological process in which ammonia, together with nitrite, is converted to dinitrogen gas according to reaction 1 [1]. Also, some nitrate is formed from nitrite. This reaction is thought to be needed for autotrophic CO2-fixation. Also, it has been suggested that CO2-fixation can be uncoupled from the catabolic reaction, i.e. the stoichiometric conversion of nitrite and ammonia to dinitrogen gas can proceed without production of cell material and nitrate [2]: Embedded Image

The bacteria, shown to be responsible for the Anammox process belong to the order of Planctomycetales. The bacteria are autotrophic and do not need organic carbon to support growth. [3] Although the bacteria are anaerobic, their activity is only reversibly inhibited by oxygen. Furthermore, Anammox bacteria are inhibited by high nitrite concentrations [4,5].

Anammox bacteria have been enriched from inocula from different wastewater treatment plants and are characterized by a low maximum growth rate, and thus have to be grown in a reactor with sufficient biomass retention [6,7]. Anammox bacteria have also been detected in several (pilot) wastewater treatment systems with high nitrogen losses and low input of organic material (COD) input [8,9].

To remove ammonia from wastewater using Anammox bacteria, these bacteria must be provided with sufficient nitrite. Nitrite can be produced from ammonia by aerobic autotrophic ammonia-oxidizing bacteria, according to reaction 2: Embedded Image

However, bacteria oxidizing ammonia to nitrite need oxygen, whereas bacteria converting ammonia and nitrite to dinitrogen gas are anaerobic. It was recently shown that both types of bacteria can co-exist in one reactor, provided that the system was kept oxygen limited. The process is called CANON, which stands for Completely Autotrophic Nitrogen removal Over Nitrite [10,11]. This process appeared to be particularly suitable for the removal of ammonia from wastewater that does not contain enough organic material to support the conventional nitrification/denitrification process [10]. Ammonia is partly oxidized to nitrite by oxygen-limited aerobic ammonia oxidizers, according to reaction 2. The nitrite produced, together with a part of the remaining ammonia, is converted to dinitrogen gas by Anammox bacteria according to reaction 1, leading to the overall reaction 3 (reaction 3): Embedded Image

In order to maintain the oxygen limitation in practice, the ammonia influx to such reactors is maintained higher than the oxygen influx [10]. In laboratory-scale CANON sequencing batch reactors, relatively low N-conversion rates have been reached until now. It was evident that the gas–liquid mass transfer of oxygen was the rate-limiting step in these reactors [10]. Gas-lift reactors are reported to have a relatively high gas–liquid mass transfer of oxygen [12]. Therefore, the current study was performed to evaluate the performance of a gas-lift reactor carrying out the Anammox and the CANON process.

2 Materials and methods

2.1 Gas-lift reactor and setup

A 1.8-l gas-lift reactor was used [13]. No special device for biomass retention was mounted or present. Synthetic wastewater (described below) was added at the top of the reactor. Gas was sparged from the bottom of the reactor at a maximum gas flow of 200 ml min−1 for fluidization of the biomass. 95% Ar/5% CO2 (Hoekloos, Schiedam, The Netherlands) and compressed air were used as gases (Table 1). When the reactor was kept anoxic, only Ar/CO2 was used. When oxygen-limited conditions were needed, Ar/CO2 mixed with air, or solely air was used. Oxygen was measured using a Clark-type electrode (Ingold, Zurich, Switzerland). The oxygen concentration was controlled by manual variation of the air supply.

View this table:
Table 1

Gas composition, oxygen concentrations in the bulk liquid and N-removal rate in the gas-lift reactor

Day no.Gas composition and flow (ml min−1)N loss (%)N-removal rate [kg N (m3 reactor)−1day−1]
  • The first 100 days, the gas-lift was operated in Anammox mode. At day 100, nitrifying biomass was added, and the mode was switched to CANON.

  • Avarage value.

  • Maximal removal.

  • Not determined.

When only CO2/Ar was used, the pH was measured but not controlled. When the gas contained air, pH was adjusted with 0.5 M Na2CO3 and 1 M HCl, using a ADI 1020 bio-controller (Applicon, Schiedam, The Netherlands).

2.2 Synthetic wastewater

Synthetic wastewater was prepared by adding ammonia and nitrite to a mineral medium in the form of NaNO2 and (NH4)2SO4. The composition of the mineral medium was (g l−1): KHCO3 1.25, KH2PO4 0.025, CaCl2·2H2O 0.3, MgSO4·7H2O 0.2, FeSO4 0.00625, EDTA 0.00625 and 1.25 ml l−1 of trace elements solution [10]. Synthetic wastewater A was used for studying the Anammox process. It contained therefore (NH4)2SO4 (6.4 g l−1) as well as NaNO2 (6.75 g l−1). To study the CANON process under oxygen-limited conditions, wastewater B was used, containing only (NH4)2SO4 (7.3 g l−1), but no nitrite.

2.3 Origin of biomass

The biomass used for inoculation with anaerobic ammonia-oxidizing bacteria originated from an Anammox sequencing batch reactor in which 80% of the biomass consisted of planctomycete-like Anammox bacteria [1]. Biomass with aerobic ammonia-oxidizing bacteria was obtained from an oxygen-limited ammonia-oxidizing sequencing batch reactor [11].

2.4 Experimental setup

The evaluation of the gas-lift reactor experiments consisted of two parts. During the first period, the gas-lift reactor was kept anoxic and during the second period, the reactor was kept oxygen limited.

In the first three months, the anoxic gas-lift reactor was used in order to grow and maintain a stable consortium of bacteria capable of Anammox. During this period biomass retained in the 20-l effluent flask was returned manually to the reactor. After the initial period, limited amounts of air were carefully introduced to support activity and growth of aerobic ammonia oxidizers. During this period, the goal was to achieve simultaneous aerobic/anaerobic ammonia oxidation. The performance of the reactor was evaluated at least every week and this was done by monitoring the ammonia, nitrate and nitrite concentrations in the influent and effluent and by subsequent calculation of the N-removal rate. Biomass was not returned to the reactor in this period.

2.5 Chemical analysis

The concentration of ammonia, nitrite and nitrate was determined colorimetrically using standard procedures [10]. Dry weight was measured after drying filtered biomass in a microwave for 10 min at 300 W.

2.6 Fluorescence in situ hybridization (FISH)

Biomass samples were fixed immediately for 2–3 h with 4% (w/v) paraformaldehyde. FISH analysis and DAPI staining was carried out as described by Juretschko et al. [14]. The 16S rRNA gene probes used were Amx820 [8], NIT3 [14], NEU [14], Nsp436 [15] and Ntspa1026 [14]. All gene probes were labelled with either of the fluorophores Cy3 and fluorescine and were purchased from Interactiva (Ulm, Germany). To estimate the relative amount of some bacterial groups (e.g. nitrite oxidizers), the DAPI-stained cells of two flocs per fixed sample were counted, as well as the fluorescently stained cells of interest.

3 Results

3.1 Ammonia removal under anoxic conditions

To investigate the possibility to grow and stably maintain Anammox biomass in a gas-lift reactor, the reactor was filled with synthetic wastewater without nitrogen compounds and was inoculated with Anammox biomass. The reactor was kept anoxic and at a pH of 7.5 by sparging with a mixture of 95% argon and 5% CO2. Accordingly, the oxygen concentration in the reactor remained always below the detection limit. After inoculation, the Anammox activity of the biomass was checked by adding ammonia and nitrite to the reactor at 14 mg N l−1 each. After 1 h, when all of the added nitrite and most of the ammonia had been converted, synthetic wastewater A with nitrite and ammonia was pumped continuously into the reactor.

At the beginning of the experiment, the superficial gas flow was relatively low, 0.003 24 m h−1 (40 ml min−1), to ensure that the biomass aggregates were not disrupted by the relatively high gas flows (and thus high shear forces) used. At this gas flow, the aggregates were fully suspended. The gas flow was then manually increased during the first days to a maximum of 0.015 m h−1 (200 ml min−1). It appeared that this had no significant effect on the activity of the Anammox biomass. Higher gas flows were not tested.

The liquid residence time at the beginning of the experiment was 43 h. During the experiment the nitrogen-loading rate was gradually increased. To increase the loading rate, the liquid residence time was gradually decreased, while the ammonia and nitrite concentration in the inflowing synthetic wastewater were kept constant. During the first part of the experiment the N-loading rate could be increased to 10.7 (±0.3) kg N (m3 reactor)−1 day−1 (Fig. 1), and at that time (days 86–100) the ammonia-consumption rate was 3.8 kg NH3–N (m3 reactor)−1 day−1, and the nitrite-consumption rate was 5.1 kg NO2–N (m3 reactor)−1 day−1. The total nitrogen-removal rate was 8.9 (±0.2) kg N (m3 reactor)−1 day−1. The liquid residence time at that moment was 6.7 h. The ratio between ammonia consumption and nitrate production was 0.2. Nitrite was present at a very low concentration of 4 (±1.5) mg l−1. The N loss was therefore as high as 80% during this part of the experiment. No aerobic ammonium and aerobic nitrite oxidizers were detectable by FISH. The biomass consisted mainly (>80%) of Anammox bacteria, which reacted with the Amx820 fluorescent probe.

Figure 1

N load (♦), liquid residence time (█) and N conversion (▲) during the anoxic period in the reactor (first part of the experiment).

3.2 Ammonia removal under oxygen-limited conditions

After 100 days of successful operation of the anaerobic gas-lift reactor, it was decided to investigate ammonia removal under oxygen-limited conditions and therefore the composition of the synthetic wastewater was changed, so that it only contained ammonia (synthetic wastewater B). The gas was then changed to a mixture of Ar/CO2 and air (Table 1), and a pH-control device was installed. In order to establish a mixed culture of bacteria capable of both aerobic and anaerobic ammonia oxidation, nitrifying biomass (0.25 g d.w. l−1) was added to the Anammox reactor. Shortly after the addition of air to the influent gas, the oxygen concentration in the bulk liquid increased to 4.4 mg l−1.

The amount of air in the gas was gradually increased (Table 1). This was done to prevent too high oxygen concentrations and to prevent too abrupt high nitrite production rates, which could lead to high nitrite concentrations in the reactor. Too high nitrite and oxygen concentrations would inhibit bacteria capable of anaerobic ammonia oxidation [16,5]. After three days, when the bulk oxygen concentration dropped from 4.4 mg l−1 to 3.7 mg l−1, indicating that the specific oxygen-consumption rate of the biomass had increased, the Ar/CO2 was removed from the influent gas and the gas consisted of only air from this moment onwards. The oxygen concentration in the reactor decreased further and after one week the oxygen concentration in the reactor stayed constant at a value of about 0.5 mg l−1, indicating that the specific oxygen-consumption rate of the biomass had now dramatically increased. The combined aerobic/anaerobic process was stable within 7 days and the N conversion stayed at 1.5 kg (m3 reactor)−1 day−1 during a period of 60 days and longer periods were not tested (Table 2).

View this table:
Table 2

Values of different parameters during steady state of the oxygen-limited period (days 111–161)

NH4+ concentration in feed (mg N l−1)1545±62
Oxygen concentration (mg l−1)0.5±0.07
NH4+ concentration (mg N l−1)899±61
NO2 concentration (mg N l−1)6±2
NO3 concentration (mg N l−1)45±7
NH4+/NO3 ratio0.07±0.01
HRT (h)10
Dilution rate (h−1)0.1
N load (kg N day−1)3.7±0.15
N conversion (kg N (m3 reactor)−1 day−1)1.5±0.2
N-removal efficiency (%)42±4.7

During this period of 60 days the process parameters were as listed in Table 2. A high excess of ammonia was observed (899 mg NH3–N l−1) and a relatively low conversion efficiency (42%). This high excess had been maintained intentionally to ensure oxygen limitation. At the start of oxygen-limited operation of the reactor, it was not known how high the maximum N-removal rate would be at a gas flow of 0.015 m h−1 (200 ml min−1). The ratio between ammonia consumption and nitrate production was 0.07. The N loss was, as expected for CANON, high and was 39% of the N load, and 93% of the N conversion.

FISH showed that the biomass consisted of mainly anaerobic Anammox bacteria reacting with probe Amx820 and aerobic ammonia oxidizers reacting to probe NEU, but a very small population of aerobic nitrite oxidizers reacting with Ntspa1026, specific for some Nitrospira species, was also present (<2%). No reaction was observed with the probes Nsp436 and NIT3 indicating the absence of Nitrosospira- and Nitrobacter-like bacteria.

4 Discussion

During the first part of the experiment, it was tried to maintain Anammox in a gas-lift reactor and to maintain a as high as possible N-conversion rate. It became clear that a relatively high ammonia-removal rate [8.9 kg N (m3 reactor)−1 day−1] can be achieved, and to our knowledge, such a high volumetric conversion rate for anaerobic ammonia oxidation has never been reported before. Compared to other reactor setups and other processes the volumetric N-removal rate is very high as well (Table 3).

View this table:
Table 3

Overview of the N conversion in kg N (m3 reactor)−1 day−1 in different reactor setups

ProcessReactorN conversionReference
Single autotrophic processes
AnammoxSBR7Strous (pers. commun.)
AnammoxGas-lift8.9This paper
Combined autotrophic processes
CANONGas-lift1.5This paper
Combined autotrophic/heterotrophic processes
  • FBR = fluidized bed reactor, SBR = sequencing batch reactor, BAS = biofilm airlift system, CSTR = continuous stirred tank reactor, N/D = nitrification/denitrification, RBC = rotating biological contactor.

  • This is the ammonium-removal rate. In all other cases, the total nitrogen removal is presented.

The nitrate production/ammonium consumption ratio was somewhat lower than expected. A ratio of 0.2 was observed, whereas a ratio of 0.3 was expected based on the stoichiometry for Anammox (reaction 1), as calculated from experiments in a sequencing batch reactor [1]. As nitrate production is thought to be coupled to biomass production, this might be an indication that the conditions in a gas-lift reactor are not as optimal for supporting growth of anaerobic ammonia oxidizers as were the conditions in a sequencing batch reactor. However, on the basis of the high N-conversion rates achieved, it is clear that a gas-lift reactor is suited to maintain and grow bacteria capable of Anammox. Probably, an even higher N-conversion rate could be achieved when a better biomass retention is applied.

During the second part of the experiment it became clear that a gas-lift reactor is also very well suited for the CANON process. The nitrate production/ammonia removal ratio was again somewhat lower than can be expected from the predicted CANON stoichiometry (reaction 3), which might be due to reduced growth of anaerobic ammonia oxidizers. Apparently, this has no effect on the stability of the process, since the process could be maintained easily for two months, and probably much longer. A very small population of aerobic nitrite oxidizers was present, but their activity must have been very low. The presence of this small population can be caused by the higher bulk oxygen concentration compared to previous studies with the CANON system with excess of ammonia [10,11]. The absence of a large and active population of nitrite oxidizers at ammonia excess is in agreement with the predictions of the model of the CANON system [17].

The major rate-limiting step was probably still the oxygen-transfer from the gas to the liquid. This can be concluded from the fact that there was a large excess of ammonia. Higher N-removal rates might be achieved when the gas–liquid oxygen transfer coefficient could be increased further. Another possibility is that the specific area of the flocs is too small to achieve a good liquid-floc mass transfer of oxygen. However, no quantification of the flocs and specific area were conducted during this experiment to address this question.

The amount of nitrifying biomass may be also a rate-limiting factor, because the oxygen concentration was low (below 0.5 mg l−1) but not zero. The amount of biomass can be increased by applying a better biomass retention. Previous experiments showed that when a sequencing batch reactor, with good biomass retention, is used to perform the CANON process, the oxygen concentration can fall below the detection limit, i.e. below 0.04 mg l−1 [10]. Nevertheless, it is confirmed that when a gas-lift reactor with very good gas–liquid transfer capabilities was used, like in this study, N-removal rates can be increased.

Compared to other setups (Table 3), a good N-conversion rate was achieved. The N conversion was 20 times higher as compared to CANON in a sequencing batch reactor, which is probably due to lower oxygen-mass-transfer rates in the sequencing batch reactor. Compared to SHARON-Anammox, CANON in a gas-lift is slightly better for removal of ammonia from high-strength wastewater streams. Moreover, CANON uses one reactor, whereas for SHARON-Anammox, two reactors are needed. In addition, the N conversion of SHARON-Anammox is limited by the maximal strength of the wastewater being treated. Compared to nitrification/denitrification, the N-removal rate of CANON is lower. However, to support denitrification, COD is needed, which is not always present in sufficient amounts in the wastewater, and addition of costly exogenous carbon sources, such as methanol, is needed.

A new ammonia-removal process has been applied in this study, with less oxygen demand and without organic carbon demand [10] in one single reactor. In this paper, it was shown that this new process is suited for treatment of high-strength wastewater. Moreover, the high nitrogen-removal capacity of this process enables compact reactor design, resulting in lower investment costs. Still, factors like maximum oxygen-transfer rates and biomass retention are good candidates for optimization.


The authors thank the ALW for funding the research (project number 80533422/P).


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
View Abstract