Chemical Phosphorus Removal with Metal Salts: Chemistry, Dosing and Sludge

What is chemical phosphorus removal and why is it used?

Chemical phosphorus removal is the precipitation of soluble phosphorus — overwhelmingly present as orthophosphate (PO₄³⁻ and its protonated forms HPO₄²⁻ and H₂PO₄⁻ at typical wastewater pH) — into an insoluble solid that is then removed with the sludge. A trivalent or divalent metal salt is dosed into the liquid, the metal cation combines with orthophosphate, and the resulting metal phosphate settles or is captured on a filter. It is the workhorse of phosphorus control on municipal and industrial plants because it is fast, robust, easy to automate and can reach far lower residuals than biology alone.

Phosphorus is the limiting nutrient in most freshwaters, so even modest discharges drive eutrophication — algal blooms, oxygen depletion and loss of ecological status. UK and EU consents for sensitive catchments now routinely demand total phosphorus below 1 mg/L, and increasingly below 0.5 or even 0.1 mg/L. Chemical dosing is the most reliable way to hit those numbers, either on its own or as a polishing stage after biological nutrient removal. Where the biological route (enhanced biological phosphorus removal, or bio-P) is exploited for the bulk load, chemical dosing typically provides the final trim and a compliance safety margin against process upsets.

The two routes differ fundamentally. Bio-P stores phosphorus inside the cells of polyphosphate-accumulating organisms and wastes it in the biomass; it is cheap in chemicals but sensitive to influent composition and redox conditions. Chemical removal converts dissolved phosphate directly into a mineral precipitate, trading chemical cost and extra sludge for reliability and precise control. Most modern works blend the two.

Which metal salts are used, and what are the reactions?

The common precipitants are trivalent iron (ferric, Fe³⁺), divalent iron (ferrous, Fe²⁺), and aluminium (Al³⁺), plus lime (Ca(OH)₂) where high-pH calcium-phosphate precipitation is acceptable. Iron and aluminium salts dominate because they work near neutral pH and integrate cleanly with activated sludge. The idealised precipitation reactions form a 1:1 metal phosphate:

The table below summarises the reagents an engineer will actually specify, with their active metal content and practical notes.

Iron generally produces a denser, faster-settling floc than aluminium, which is why iron-salt dosing for wastewater treatment is the default on most activated-sludge plants. Aluminium yields a lighter, more voluminous hydroxide floc but avoids the reddish colour that ferric residuals can impart.

Two mechanistic subtleties matter when specifying a reagent. First, the precipitate is rarely pure stoichiometric FePO₄ or AlPO₄; it is an amorphous, hydrated, mixed-valence solid whose composition drifts with pH, mixing intensity and ageing. Freshly formed floc adsorbs additional phosphate onto its high surface area, then slowly recrystallises to a lower-surface-area, less reactive form — which is why rapid, energetic mixing at the dosing point improves efficiency and why aged or recycled floc removes less phosphate per unit metal. Second, the counter-ion is not inert: chloride from ferric chloride adds to effluent salinity, while the sulphate from ferric or alum can feed sulphate-reducing bacteria in downstream anaerobic zones and digesters, so the anion is a genuine design variable, not a detail.

Why does the real dose exceed the 1:1 stoichiometry?

This is the single most important concept in phosphorus dosing, and where the theory bites. The clean 1:1 reaction above never governs the real dose. Two effects force the ratio up:

1. Competing hydroxide precipitation. Fe³⁺ and Al³⁺ are strong Lewis acids that hydrolyse in water. A large fraction of the dosed metal precipitates as ferric or aluminium hydroxide (Fe(OH)₃, Al(OH)₃) rather than reacting with phosphate:

2. Solubility equilibrium at low residual P. FePO₄ and AlPO₄ have finite solubility products. As the target residual phosphate falls, Le Chatelier’s principle requires a rising excess of free metal ion to push the equilibrium and suppress the dissolved phosphate concentration. The consequence is a dose–residual relationship that is mildly super-stoichiometric at high effluent P but rises steeply and non-linearly as the target approaches zero. Reaching <0.1 mg/L can demand molar Me:P ratios of 3:1 to 5:1 or higher — several times the ideal — because you are chasing an asymptote.

These are indicative ranges — the exact curve is effluent-specific and must be fixed by jar tests or plant trials. The key engineering lesson is that halving the phosphorus target roughly doubles the incremental metal demand near the low end, with corresponding cost and sludge penalties.

Worked example: metal dose from a molar ratio

Consider a works treating 10,000 m³/d with a soluble phosphorus load to be precipitated of 6 mg/L as P, aiming to knock this down to about 1 mg/L. We will dose ferric chloride at a design molar ratio of 2.0 mol Fe : 1 mol P. Carry the numbers through:

• P to remove: (6 − 1) mg/L × 10,000 m³/d = 5 g/m³ × 10,000 = 50,000 g/d = 50 kg P/d.
• Moles of P: molar mass of P = 31 g/mol, so 50,000 / 31 = 1,613 mol P/d.
• Moles of Fe: at 2.0:1, that is 2 × 1,613 = 3,226 mol Fe/d.
• Mass of Fe: molar mass Fe = 55.8 g/mol, so 3,226 × 55.8 = 180,000 g/d = 180 kg Fe/d.
• As FeCl₃: molar mass 162.2 g/mol, so 3,226 × 162.2 = 523,000 g/d = 523 kg FeCl₃/d (100% basis).
• As delivered 40% solution (density ~1.4 kg/L): 523 / 0.40 = 1,308 kg product/d ÷ 1.4 = ~935 L/d of ferric chloride solution.

So this plant needs on the order of 0.9 m³/d of 40% ferric to hold 1 mg/L. That single figure — not the elegant 1:1 equation — is what sizes the storage tank, dosing pump and chemical budget.

How does the dose change as the target P falls?

To make the dose-ratio sensitivity concrete, take the same 10,000 m³/d works with 6 mg/L soluble P in, and tighten the consent progressively. Using representative molar ratios from the table above, the ferric demand behaves like this:

• Target 1.0 mg/L — remove 5 mg/L, ratio ~2.0:1 → 180 kg Fe/d (baseline).
• Target 0.5 mg/L — remove 5.5 mg/L, ratio ~3.0:1 → moles P = 55,000/31 = 1,774; Fe = 5,323 mol = 297 kg Fe/d. That is +65% metal for a 0.5 mg/L tightening.
• Target 0.1 mg/L — remove 5.9 mg/L, ratio ~4.5:1 → moles P = 59,000/31 = 1,903; Fe = 8,565 mol = 478 kg Fe/d. That is +165% versus baseline — the metal demand has nearly tripled to shave the last 0.9 mg/L.

The non-linearity is stark: the marginal cost of each additional increment of phosphorus removed rises sharply near the compliance floor. This is exactly why plants chasing very stringent limits favour a two-stage strategy — a modest co-precipitation dose upstream to do the bulk work economically, followed by a small, highly effective tertiary dose plus filtration for the final polish. Trying to reach 0.1 mg/L with a single upstream dose is chemically wasteful and generates far more sludge than a split scheme. Where the ultra-low target drives the design, it is worth comparing the whole-life chemical and sludge cost against nutrient-recovery options and confirming the discharge requirement with the applicable effluent discharge standards before committing.

Where should the metal salt be dosed?

Dosing point is as important as dose. There are three classic locations, and many plants use more than one:

• Primary / pre-precipitation — dosed ahead of primary settlement. Removes phosphorus and enhances primary solids capture, and the precipitate leaves in primary sludge. Simple and reduces load on the biology, but reagent efficiency is lower and it strips substrate that bio-P and denitrification may need downstream.
• Simultaneous / co-precipitation — dosed directly into the aeration tank (or mixed liquor) of an activated-sludge plant. This is the most common arrangement: the metal precipitates within the floc and leaves with the waste-activated sludge. Excellent contact and good reagent use, but it consumes alkalinity in the reactor and adds inert solids to the mixed liquor, lowering the volatile fraction and effectively raising the MLSS the clarifiers must handle.
• Tertiary / post-precipitation — dosed after secondary clarification, followed by a dedicated separation stage (tertiary clarifier, sand filter, disc filter or membrane). This delivers the lowest residuals — reliably <0.1–0.5 mg/L — because the precipitate is captured in a clean, dedicated step, but it needs extra capital plant and produces a separate chemical sludge stream.

The trade-off is efficiency versus attainable residual. Primary and simultaneous dosing are cheap and integrate with existing plant but plateau around 0.5–1 mg/L. Tertiary dosing with filtration is the only route to consistent sub-0.1 mg/L performance because it removes the fine metal-phosphate particulates that would otherwise wash over a clarifier weir. A well-designed scheme frequently pairs a co-precipitation dose (bulk removal) with a small tertiary trim, which minimises total reagent while guaranteeing the consent.

What extra sludge does chemical dosing produce?

Every mole of metal dosed ends up as solids — either as metal phosphate or as metal hydroxide — so chemical phosphorus removal unavoidably increases sludge production. This is a real operating cost: more sludge to thicken, dewater, transport and dispose of, and a lower volatile fraction that can affect digestion and cake quality. Estimating it is straightforward if you track both product phases.

Worked example. Using the baseline case above — 50 kg P/d removed (1,613 mol P/d) at a 2.0:1 molar ratio (3,226 mol Fe/d):

• FePO₄ formed: 1 mol Fe pairs with each mol P → 1,613 mol FePO₄ × 150.8 g/mol = 243,000 g/d = 243 kg/d.
• Excess Fe as hydroxide: 3,226 − 1,613 = 1,613 mol Fe → Fe(OH)₃ = 1,613 × 106.9 g/mol = 172,000 g/d = 172 kg/d.
• Total extra dry chemical sludge: 243 + 172 = ~415 kg DS/d.

That is a substantial addition — for context, comparable in magnitude to the primary solids of a small works — and it scales with the metal excess. Push the ratio to 4.5:1 for a 0.1 mg/L target and the hydroxide fraction roughly triples, so the sludge penalty is another compelling reason not to over-dose. The upside is that phosphorus locked into sludge can be recovered: interest in struvite phosphorus recovery is partly driven by the desire to reclaim this nutrient rather than landfill it, although iron-bound phosphate is less amenable to struvite recovery than the phosphorus released in bio-P digestion.

What about alkalinity and pH?

Metal-salt dosing is acidic. Each mole of Fe³⁺ or Al³⁺ that hydrolyses releases up to three moles of H⁺, consuming bicarbonate alkalinity. As a rule of thumb, dosing ferric or alum destroys roughly 2.7–5.4 g of alkalinity (as CaCO₃) per gram of metal, depending on how much hydrolyses. On a plant with low natural alkalinity this can depress pH below the optimum for both precipitation and nitrification, so alkalinity must be monitored and, if necessary, supplemented with lime, caustic or sodium bicarbonate.

pH also sets the precipitation optimum. Aluminium phosphate is least soluble around pH 5.5–6.5, and ferric phosphate around pH 5–7; both metals shift toward hydroxide formation as pH rises above about 7–8, which reduces phosphate-specific efficiency (though co-precipitation still captures phosphate onto the hydroxide). In an activated-sludge reactor held near pH 7 for nitrification you are therefore dosing at a compromise pH, which is another reason the real ratio exceeds the ideal. Sodium aluminate is sometimes chosen precisely because it adds alkalinity instead of consuming it, buffering the pH swing where alkalinity is scarce.

How is the dose controlled in practice?

A fixed dose set to the worst-case load is simple but wasteful, because phosphorus load varies diurnally and with flow. Three control philosophies are used, in ascending order of sophistication.

• Flow-proportional (open loop). The dosing pump is paced to plant flow at a fixed metal-to-flow ratio. Cheap and stable, but blind to changes in influent phosphorus concentration, so it must be set with a safety margin and therefore over-doses much of the time.
• Feed-forward on load. An online orthophosphate analyser upstream, combined with the flow signal, computes the actual phosphorus load and trims the dose to a target ratio. This tracks the diurnal load curve and cuts average chemical use appreciably.
• Feedback (closed loop). An effluent phosphate analyser after the separation stage trims the dose to hold a residual set-point, usually cascaded with a feed-forward term so the loop is not chasing the long hydraulic lag of the plant on its own.

Because the dose–residual curve is so non-linear near the limit, closed-loop control pays for itself fastest on plants with tight consents: a small error in dose translates into a large error in effluent phosphorus at low residuals, so instrumentation that keeps the plant on the flat part of the cost curve avoids both non-compliance and gross chemical over-spend. Reliable phosphate analysers, regular reagent-strength verification and disciplined pump calibration are therefore not optional extras but core to the economics of chemical phosphorus removal. Where a works must demonstrate compliance against a numeric permit, the control scheme and its sampling regime should be designed hand-in-hand with the monitoring obligations rather than bolted on afterwards.