Nitrate Removal From Water: Ion Exchange, Denitrification and RO

Why is nitrate so hard to remove from water?

Nitrate is chemically one of the most awkward contaminants in water treatment, and understanding why explains every removal technology that follows. Three properties combine against the engineer:

• It is extremely soluble. Alkali and alkaline-earth nitrates are all freely soluble in water, so — unlike phosphate, carbonate or metals — nitrate cannot be precipitated out by dosing a coagulant, adjusting pH or adding a counter-ion. There is no chemical clarification route.
• It is non-volatile. Nitrate has no vapour pressure of its own, so air stripping, steam stripping and thermal degassing — effective for ammonia, CO₂ or VOCs — do nothing. The nitrogen has to be either transferred to a solid phase or chemically reduced to a gas.
• It is a small, singly-charged anion. Nitrate is monovalent, weakly hydrated and chemically inert under ambient redox conditions. It does not adsorb strongly onto activated carbon or conventional media, and it is stable — it will not oxidise or reduce on its own timescale relevant to a treatment works.

The consequence is that only two physical strategies exist. Either you separate the nitrate ion from the water (ion exchange transfers it onto a resin; reverse osmosis and electrodialysis reject or migrate it across a membrane), or you destroy it by biologically reducing nitrogen from the +5 oxidation state in NO₃^- all the way to elementary dinitrogen gas (N₂), which is inert and leaves as a harmless off-gas. Separation always produces a concentrated nitrate-bearing waste stream that still needs a home; destruction avoids that but demands an electron donor and a living biomass to maintain. That fundamental split — concentrate-and-dispose versus reduce-and-vent — is the decision at the heart of every nitrate project.

What are the health limits and regulatory drivers?

The reason nitrate is regulated so tightly in drinking water is a condition called methaemoglobinaemia, historically known as blue-baby syndrome. Ingested nitrate is reduced to nitrite (NO₂^-) — in bottle-fed infants under about six months, by bacteria in the less-acidic infant gut. Nitrite then oxidises the ferrous iron (Fe²⁺) at the centre of haemoglobin to the ferric state (Fe³⁺), forming methaemoglobin, which cannot bind and transport oxygen. Above roughly 10% methaemoglobin the infant becomes cyanotic; the effect is acute rather than chronic, which is why the limit is protective of the most vulnerable group.

The numbers you must design to:

• WHO guideline and EU / UK drinking-water limit: 50 mg/L as NO₃^-, which equals 11.3 mg/L expressed as nitrogen (NO₃-N). The conversion is simply the mass ratio of nitrogen to nitrate, 14/62 = 0.226.
• US EPA Maximum Contaminant Level: 10 mg/L as NO₃-N (equivalent to 44.3 mg/L as NO₃^-), with a separate 1 mg/L limit for nitrite as N.

Always confirm whether a specification, laboratory report or consent is stated as nitrate or as nitrogen — mixing the two introduces a factor-of-4.4 error that has ruined more than one mass balance. For discharges to the environment, nitrate is regulated as part of total nitrogen under nutrient and eutrophication controls rather than the health-based drinking limit; you can review how those effluent thresholds are framed in our guide to nitrogen discharge and consent limits, and in the sibling article on effluent discharge standards in the UK.

How does ion exchange remove nitrate?

Ion exchange (IX) is the most common route for point-of-entry and municipal nitrate removal from water. A bed of strong-base anion (SBA) resin is operated in the chloride form; as water passes through, nitrate ions are picked up onto the resin and an equivalent amount of chloride is released:

R–Cl + NO₃^- → R–NO₃ + Cl^-

The difficulty is selectivity. The equilibrium between two anions competing for the resin is described by a selectivity coefficient:

Because sulphate out-competes nitrate on a conventional resin, a dangerous failure mode appears: chromatographic nitrate dumping (peaking). As the bed runs, sulphate progressively displaces nitrate that was loaded earlier, pushing a concentrated band of nitrate down the column. Near exhaustion the treated water can briefly contain more nitrate than the raw feed — a compliance disaster if the run is not stopped in time. The engineering answer is a nitrate-selective resin: quaternary-ammonium sites built with larger, more hydrophobic alkyl groups (triethyl- or tributyl-amine functionality) that reverse the order to NO₃^- > SO₄^2-, suppressing dumping at the cost of slightly lower total capacity.

The exhausted resin is regenerated with a concentrated sodium chloride brine (typically 8–12% NaCl), which mass-action drives the nitrate back off and returns the resin to the chloride form. The mechanics of the resin bed, service cycle and regeneration are shared with softening; the fundamentals are covered in ion exchange water softening.

Worked example: sizing a nitrate ion-exchange bed

Consider a small groundwater supply that must be de-nitrated:

• Flow Q = 40 m³/h; nitrate influent 80 mg/L as NO₃^- (= 18.1 mg/L as N), target < 50 mg/L.
• Competing sulphate 60 mg/L as SO₄^2-; bicarbonate is weakly held and largely passes.
• Nitrate-selective resin, operating capacity 0.9 eq/L (900 meq per litre of resin).

Step 1 — ionic load. Convert each competing anion to milli-equivalents per litre using its equivalent weight (NO₃^- = 62 g/eq; SO₄^2- = 48 g/eq):

• Nitrate: 80 / 62 = 1.29 meq/L
• Sulphate: 60 / 48 = 1.25 meq/L
• Total anions loading the resin ≈ 2.54 meq/L

Step 2 — throughput per litre of resin. Dividing resin capacity by the ionic load gives the volume treated before exhaustion:

900 meq/L ÷ 2.54 meq/L = 354 bed volumes per cycle.

Step 3 — resin volume for a 24 h run. At 40 m³/h, a 24-hour service run treats 960 m³. The resin volume needed is:

V = 960,000 L ÷ 354 = 2,712 L ≈ 2.7 m³ of resin.

This would be split across two vessels (duty/regenerating) so supply is continuous. The empty-bed contact time at this volume is 2.7 / 40 × 60 ≈ 4 minutes, comfortably within the 1.5–5 minute range that anion resins need.

Step 4 — regenerant brine. At a regeneration dose of 120 g NaCl per litre of resin, each regeneration consumes 2,700 L × 120 g/L = 324 kg NaCl. Delivered as 10% brine that is about 3.2 m³ of spent regenerant per cycle, laden with the stripped nitrate, sulphate and surplus salt. That brine — not the treated water — is the real project constraint, because its disposal is often harder and costlier than the treatment itself.

How does biological denitrification remove nitrate?

Biological denitrification takes the opposite approach: instead of concentrating nitrate, it destroys it. Under anoxic conditions (dissolved oxygen near zero, nitrate present) facultative bacteria use nitrate as a terminal electron acceptor in place of oxygen, reducing it stepwise NO₃^- → NO₂^- → NO → N₂O → N₂. The nitrogen leaves as inert gas, so there is no nitrate-bearing liquid waste — the great advantage of the biological route.

Every electron acceptor needs an electron donor, and the choice of donor defines the two families:

• Heterotrophic denitrification uses an organic carbon source. With methanol the McCarty half-reaction synthesis-and-energy balance is approximately:
  NO₃^- + 1.08 CH₃OH + 0.24 H₂CO₃ → 0.056 C₅H₇O₂N + 0.47 N₂ + 1.68 H₂O + HCO₃^-
• Autotrophic denitrification uses an inorganic donor and needs no added carbon. Hydrogenotrophic bacteria use hydrogen (2 NO₃^- + 5 H₂ → N₂ + 2 OH^- + 4 H₂O); sulphur-based systems use elemental sulphur or thiosulphate via Thiobacillus denitrificans, which is attractive because no dosing is required but which generates sulphate and consumes alkalinity strongly — you can trade a nitrate problem for a sulphate-and-pH problem if the raw water is soft.

The carbon (or hydrogen) demand is not optional overhead — it is fixed by stoichiometry, and it also has to satisfy any dissolved oxygen present, which the biomass will always consume in preference to nitrate. The governing dose relationship for methanol is:

For a fuller treatment of anoxic-zone design, recycle ratios and the nitrification–denitrification balance, see our sibling article on biological nutrient removal.

Worked example: carbon dose for denitrification

Take an effluent-polishing duty at the same 40 m³/h flow, reducing nitrate nitrogen from 18 mg/L to a target of 2 mg/L, with 2 mg/L of dissolved oxygen carried into the anoxic zone.

Step 1 — nitrogen to be reduced. Δ(NO₃-N) = 18 − 2 = 16 mg/L.

Step 2 — apply the McCarty dose. With negligible nitrite:

C_m = 2.47 × 16 + 0.87 × 2 = 39.5 + 1.7 = 41.2 mg/L methanol.

Step 3 — mass dosing rate. At 40 m³/h:

41.2 g/m³ × 40 m³/h = 1,648 g/h ≈ 1.65 kg/h methanol, or about 40 kg/day.

Step 4 — sanity checks. Methanol carries roughly 1.5 g COD per g, so the dose adds about 62 mg/L of COD that the denitrifiers must fully consume; over-dosing leaves residual BOD/COD in the treated water, which is why methanol feed is trimmed to the measured nitrate load with a modest safety margin rather than run rich. The reaction also returns roughly 3.57 × 16 ≈ 57 mg/L of alkalinity as CaCO₃, a useful buffer where upstream nitrification has depressed pH. Compared with the ion-exchange case, the running cost here is a continuous chemical (methanol) rather than a periodic brine, and the only solid waste is a small quantity of surplus biomass.

When is reverse osmosis or electrodialysis the answer?

Membrane processes remove nitrate as part of a broader demineralisation, and they earn their place where the water needs several contaminants stripped at once (hardness, sulphate, total dissolved solids, nitrate) or where a very low nitrate residual is required. Two options:

• Reverse osmosis (RO) rejects nitrate by size and charge exclusion. Because nitrate is a small monovalent ion, its rejection (typically 85–95%) is lower than that of divalent sulphate (> 99%), so nitrate is one of the harder ions for RO to hold. Design and staging are covered in reverse osmosis system design.
• Electrodialysis reversal (EDR) pulls ions across ion-selective membranes under a DC field and periodically reverses polarity to self-clean; it can be more robust than RO on scaling waters and recovers a higher fraction of the feed.

Worked RO mass balance. Feed 80 mg/L NO₃^-, observed rejection R = 0.90, system recovery r = 0.75.

• Permeate: C_p = (1 − R) × C_f = 0.10 × 80 = 8 mg/L — comfortably below the 50 mg/L limit.
• Concentrate by mass balance: C_c = (C_f − r·C_p) / (1 − r) = (80 − 0.75×8) / 0.25 = 296 mg/L NO₃^-, in a reject stream that is 25% of the feed flow.

So a membrane plant produces excellent permeate but exports a concentrated nitrate brine at roughly four times the feed concentration — the same concentrate-and-dispose problem as ion exchange, just in a different package. A blended-permeate design (treating only part of the flow and blending back raw water to just under the limit) cuts both membrane area and reject volume, and is standard practice for nitrate duty.

How do you compare the technologies and manage the waste?

The choice between the three routes is usually decided not by the treated-water quality — all can meet 50 mg/L — but by the waste stream each leaves behind and the local disposal options for it.

Managing the residuals is where projects succeed or fail:

• Brine and RO concentrate are the dominant issue for the separation routes. Disposal options include sewer discharge (subject to a trade-effluent consent and often a salinity or nitrogen cap), evaporation ponds in dry climates, or further concentration towards zero liquid discharge. Because the brine carries the whole nitrate load in a small volume, its nitrogen mass has simply been relocated, not removed — the regulator will hold you to account for that mass.
• Biomass from denitrification is modest (the cell yield on nitrate is low) and is handled as ordinary waste sludge; the bigger operational care is dosing control to avoid leaving residual carbon in the product water.
• Hybrid schemes are common: ion exchange or RO to hit the number reliably, with the concentrated reject sent to a compact biological denitrification stage that destroys the nitrate and cuts the mass sent to disposal — capturing the reliability of separation and the mass-destruction benefit of biology in one train.

Where the treated water feeds a potable or high-purity duty and a further membrane polishing stage is planned, the nitrate strategy should be chosen alongside it rather than bolted on afterwards, so that reject streams and blend ratios are optimised as a whole.

How to select and size a nitrate removal process

1. Confirm the target basis. Establish whether the limit is as NO3 or as N (50 mg/L NO3 = 11.3 mg/L N) and the required residual, then characterise flow, nitrate, sulphate, alkalinity and temperature.
2. Decide separate or destroy. Choose between concentrating nitrate (ion exchange, RO/EDR) and destroying it biologically, based on the availability of a brine/reject disposal route.
3. For ion exchange, specify a nitrate-selective resin. Use a nitrate-selective SBA resin to reverse the sulphate-over-nitrate order and prevent chromatographic nitrate dumping; size bed volumes from the total competing anion load.
4. For denitrification, fix the carbon dose. Apply the McCarty relationship (Cm = 2.47 NO3-N + 1.53 NO2-N + 0.87 DO) to set the methanol or alternative donor dose, and check the alkalinity balance.
5. For membranes, run the mass balance. Compute permeate and concentrate nitrate from rejection and recovery; consider a blended-permeate design to cut membrane area and reject volume.
6. Plan the waste stream first. Confirm the disposal route for spent brine, RO concentrate or surplus biomass before finalising the process, since residual management usually governs cost.