Anammox Deammonification: The Science of PN/A Nitrogen Removal

What is anammox deammonification?

Deammonification is autotrophic nitrogen removal built on two coupled reactions. First, aerobic ammonia-oxidising bacteria (AOB) convert about half of the incoming ammonium to nitrite — this is partial nitritation. Second, anammox bacteria anaerobically oxidise the remaining ammonium using that nitrite as the electron acceptor, releasing dinitrogen gas. Together the two steps are abbreviated PN/A (partial nitritation/anammox), and the combined process is what the water industry calls deammonification.

The engineering appeal is that it short-circuits the classical nitrogen cycle used in activated sludge. Conventional treatment oxidises ammonium all the way to nitrate (full nitrification), then reduces nitrate back to nitrogen gas using an organic electron donor (denitrification). Deammonification skips the energetically expensive nitrite-to-nitrate step and the carbon-hungry denitrification step entirely. Because anammox bacteria fix their own carbon from bicarbonate, no methanol, acetate or influent BOD is consumed to drive the reaction.

The technology first went to full scale on sidestream flows — the warm, ammonium-rich liquor returned when digested sludge is dewatered — and is now the reference process for that duty. Extending it to the cold, dilute mainstream is the central research frontier in municipal nitrogen removal. This article works through the microbiology, the governing stoichiometry, the quantified savings, and the reactor-design consequences of the anammox reaction's notoriously slow growth.

How do anammox bacteria remove nitrogen?

The anammox organisms are a deep-branching group of the phylum Planctomycetes — genera such as Candidatus Brocadia, Kuenenia and Anammoxoglobus. They are chemolithoautotrophs: they draw energy from oxidising ammonium and fix inorganic carbon for growth. Uniquely, they carry out the reaction inside a dedicated intracytoplasmic organelle, the anammoxosome, whose ladder-like ladderane lipid membrane is dense enough to contain the toxic intermediates and maintain the proton-motive force.

Mechanistically, nitrite is first reduced to nitric oxide, which is then combined with ammonium to form hydrazine (N₂H₄) by the enzyme hydrazine synthase. Hydrazine — the same compound used as rocket fuel — is a powerful reductant; its oxidation to N₂ by hydrazine dehydrogenase yields the four electrons that regenerate nitric oxide and drive carbon fixation. The catabolic reaction, as first quantified by Strous and co-workers, is:

Two numbers from this stoichiometry govern everything downstream. The 1.32:1 nitrite-to-ammonium ratio tells the process designer exactly how much of the influent ammonium must be nitritated: about 57% (1.32 / 2.32), which is why partial nitritation is targeted at "roughly half". And the tiny biomass coefficient — a growth yield near 0.11 g VSS per g N removed — explains both the low sludge production and the painfully slow growth: doubling times of 10–14 days at 30–35 °C, an order of magnitude slower than ordinary heterotrophs. Retaining this biomass, as we discuss below, is the whole game.

What is partial nitritation and why suppress NOB?

Partial nitritation is the aerobic front half of deammonification. Ammonia-oxidising bacteria convert ammonium to nitrite via the reaction NH₄⁺ + 1.5 O₂ → NO₂^- + 2 H⁺ + H₂O. The design intent is to stop there — to produce a roughly 1:1.3 mixture of ammonium and nitrite feeding the anammox step — and crucially to prevent the second nitrification stage, nitratation, in which nitrite-oxidising bacteria (NOB) such as Nitrobacter and Nitrospira convert nitrite onward to nitrate.

NOB are the process's principal enemy for two reasons. Every gram of nitrite they oxidise to nitrate is a gram stolen from the anammox bacteria, and the extra oxidation wastes oxygen. Suppressing NOB while retaining AOB and anammox is therefore the core control problem. Four levers are used, usually in combination:

• Dissolved oxygen. AOB have a higher affinity for oxygen than NOB, so holding DO low (typically 0.2–1.0 mg/L) lets AOB out-compete NOB for the limited oxygen.
• Temperature and SRT. Above about 25 °C the maximum growth rate of AOB exceeds that of NOB. Operating at a deliberately short aerobic solids retention time then washes NOB out faster than they can grow, while the faster AOB are retained — the classic Hellinga "SHARON" strategy.
• Free ammonia and free nitrous acid. At the high pH and ammonium of digester liquor, free ammonia (NH₃) inhibits NOB more strongly than AOB (the Anthonisen thresholds), giving a selective handicap.
• Transient anoxia / intermittent aeration. NOB adapt sluggishly to repeated anoxic-aerobic swings, so cycling aeration favours AOB and anammox.

A neat piece of chemistry makes sidestream partial nitritation partly self-regulating. Nitritation releases 2 mol H⁺ per mol N and so consumes about 7.14 g alkalinity as CaCO₃ per g N nitrited. Digester dewatering liquor carries roughly one mole of bicarbonate per mole of ammonium — only about half of what full nitritation would need. The reaction therefore runs out of alkalinity after converting close to half the ammonium, and pH falls, naturally arresting nitritation at the 50–57% needed to feed anammox. For a deeper treatment of the wider nitrogen and phosphorus removal context, see our overview of biological nutrient removal.

How much energy, carbon and sludge does it save?

The savings are large and quantifiable, and they all flow from skipping nitratation and denitrification. Consider the oxygen requirement first. Full nitrification of ammonium to nitrate needs the theoretical 4.57 g O₂ per g N (3.43 for nitritation plus 1.14 for nitratation). Deammonification only nitritates about 57% of the nitrogen and does so only to nitrite:

Worked example: sizing a sidestream deammonification reactor

Take a realistic sidestream duty: dewatering centrate at 50 m³/h carrying 1,000 mg/L NH₄-N, arriving warm at 32 °C. Work the nitrogen load, the savings and the reactor volume from first principles.

• Nitrogen load: 50 m³/h × 1,000 g/m³ = 50,000 g N/h = 50 kg N/h = 1,200 kg N/d.
• Oxygen — conventional: 4.57 g O₂/g N × 50 kg N/h = 228.5 kg O₂/h.
• Oxygen — PN/A: 1.9 g O₂/g N × 50 kg N/h = 95 kg O₂/h, a saving of 133.5 kg O₂/h (−58%). At an aeration efficiency of ~1.5 kg O₂/kWh that is roughly 89 kW, or about 2,100 kWh/d, avoided.
• External carbon avoided: conventional denitrification of 1,200 kg N/d at ~3 g methanol/g N would demand ~3,600 kg methanol/d; deammonification needs none.
• Reactor volume: a single-stage granular/DEMON reactor sized at a volumetric removal rate of 1.0 kg N/m³·d needs 1,200 / 1.0 = 1,200 m³. Pushed to a well-retained 2.0 kg N/m³·d it halves to ~600 m³.
• Sludge production: at a combined yield of 0.12 g VSS/g N, biomass output is 0.12 × 1,200 = ~144 kg VSS/d, roughly a tenth of a carbon-driven denitrification lane on the same load.

The alkalinity check confirms the process is feasible without dosing. Partial nitritation of ~57% of 1,200 kg N/d consumes 0.57 × 7.14 ≈ 4.1 g CaCO₃/g N, i.e. ~4,900 kg CaCO₃/d. Centrate at this ammonium concentration typically supplies ~3,000–3,600 mg/L alkalinity as CaCO₃, which is close to sufficient; any shortfall is trimmed with a modest bicarbonate dose rather than the tonnage of methanol the conventional route would demand. These first-pass figures should always be firmed up through a proper process design and reactor-selection study with pilot data before procurement.

Single-stage versus two-stage: which configuration?

Deammonification is delivered in two architectures. In the two-stage layout the partial nitritation happens in a dedicated aerobic reactor (the original SHARON concept) and the effluent, now a balanced ammonium/nitrite mixture, feeds a separate anaerobic anammox reactor. In the single-stage layout both AOB and anammox live in one tank under low, often intermittent, aeration — with the anammox held in granules, on carriers, or in flocs while the AOB occupy the aerobic outer shell of the same aggregates. Commercial single-stage names include DEMON (a pH-controlled SBR that uses a hydrocyclone to selectively retain heavy anammox granules and waste light flocs), ANAMMOX granular reactors, and MBBR/IFAS carrier systems such as ANITA Mox.

In the single-stage tank the low bulk oxygen keeps the granule interior anoxic, so nitrite made at the surface is consumed by anammox before it can accumulate to inhibitory levels — an elegant self-protection that also makes NOB suppression trickier, because AOB and anammox must share one dissolved-oxygen set-point. The single-stage approach dominates the sidestream market on footprint and simplicity; two-stage designs retain a niche where very tight, independently tunable control of nitritation is wanted, or where the two populations are best decoupled. Most modern sidestream retrofits choose single-stage.

Why is mainstream anammox so difficult?

If deammonification is so efficient on sidestream, why not run the entire plant on it? The prize is enormous — a mainstream plant that removed nitrogen autotrophically could approach energy neutrality, or even net energy positive, by diverting influent carbon to the digesters for gas rather than burning it in aeration. But three physical realities make mainstream anammox genuinely hard.

• Temperature. Anammox thrives at 30–40 °C, exactly the temperature of digester liquor. Municipal mainstream sits at 10–20 °C, where the anammox growth rate falls steeply and the doubling time can stretch past a month, so biomass retention becomes even more critical.
• NOB suppression collapses. The temperature-and-SRT trick that washes NOB out of a warm sidestream reactor fails in the cold, because below ~15 °C NOB actually grow faster than AOB. Holding NOB down in cold mainstream conditions is the single hardest problem, addressed with tight low-DO control, intermittent aeration and residual-ammonium set-points.
• Low, variable concentration. Mainstream ammonium is 30–60 mg/L, not 1,000, so reaction rates are lower and the driving force is weaker; anammox must also compete with ordinary heterotrophs for the nitrite when influent BOD is present.

Pilot and demonstration plants — often coupling an A-stage carbon-capture contact stabilisation step ahead of a B-stage PN/A polishing step — have shown mainstream deammonification is possible, but reliable year-round NOB-out control at low temperature remains the barrier to routine full-scale adoption. It is one of the most active areas in municipal wastewater research, and success would reset the energy economics of the whole sector, complementing resource-recovery advances such as struvite phosphorus recovery from the same sidestream liquor.

How do you retain the slow-growing biomass?

Everything about deammonification design bends around one fact: anammox bacteria grow at doubling times of 10–14 days when warm, far longer when cold. A conventional flocculent activated-sludge SRT of 5–15 days would simply wash them out. The reactor solids retention time must comfortably exceed the anammox doubling time — usually a target well above 20–30 days — and this is achieved by physically decoupling biomass retention from hydraulic retention. Three strategies dominate:

• Granulation. Anammox self-aggregate into dense, fast-settling granules (settling velocities many times a floc's). DEMON exploits this directly, running the mixed liquor through a hydrocyclone that keeps the heavy granules and discards the light, NOB-laden flocs — a continuous selection pressure that both retains anammox and helps suppress NOB.
• Carriers (MBBR/IFAS). A biofilm on plastic carriers grows a redox-stratified community — AOB in the aerobic outer layer, anammox in the anoxic core — and the carriers are screened back into the tank, so biomass age is independent of hydraulic residence time.
• Membrane or settler retention. Membranes or lamella settlers give near-total biomass capture for the most demanding, low-temperature mainstream duties.

Two operational sensitivities deserve constant vigilance. Anammox is inhibited by its own substrate nitrite above roughly 100 mg NO₂-N/L for sustained exposure, so partial nitritation must never over-run — another argument for the single-stage layout, where nitrite is consumed as fast as it forms. And because start-up means growing a very slow organism from a small seed, commissioning can take months; the industry now shortcuts this by seeding new reactors with granules or biomass from an established plant, turning a six-month start-up into a few weeks. Get retention, nitrite control and seeding right, and deammonification is a robust, low-energy workhorse for concentrated ammonium streams. For a broader tour of the nitrogen-removal options it competes with, see our guide to industrial wastewater treatment processes.

How to specify a sidestream deammonification process

1. Characterise the sidestream. Measure centrate flow, NH4-N, alkalinity and temperature. Confirm the stream is warm (>25 C) and carbon-poor so PN/A is favoured.
2. Compute the nitrogen load and target. Calculate kg N/d and the ~57% partial-nitritation fraction from the 1.32:1 anammox nitrite:ammonium ratio.
3. Choose single- or two-stage. Pick single-stage (DEMON/granular/MBBR) for lowest footprint and in-situ nitrite control; two-stage where independent nitritation control is required.
4. Size the reactor volume. Divide the nitrogen load by a volumetric removal rate of 0.5-2 kg N/m3.d, allowing margin for the slow biomass and winter turndown.
5. Design NOB suppression and retention. Set low DO (0.2-1.0 mg/L), short aerobic SRT and biomass retention (hydrocyclone, carriers) so NOB wash out while anammox are held above their doubling time.
6. Plan seeding and start-up. Seed with granules or biomass from an operating plant to shorten commissioning from months to weeks, and monitor nitrite to stay below ~100 mg N/L.