The Short Answer
No. A conventional ultrafiltration (UF) membrane does not remove dissolved polyphosphate. UF separates materials by physical size, and dissolved polyphosphate chains are far smaller than the pores of a standard UF membrane. If polyphosphate is present in the UF feed water in a truly dissolved form, it will pass into the permeate essentially unchanged.
There are narrow exceptions. A very “tight” UF membrane with a low molecular weight cut-off, a polyphosphate fraction made of unusually long chains, or polyphosphate that has already associated with calcium, iron, aluminium, or organic colloids can be partially retained. Those cases are conditional, and they are constantly confused with genuine dissolved-phosphate removal. The rest of this article explains where the boundary sits and how to check it in your own system.
What Ultrafiltration Actually Rejects
UF sits between microfiltration and nanofiltration in the pressure-driven membrane family. Its job is clarification and disinfection, not desalination.
- Typical nominal pore size: roughly 0.01–0.1 µm.
- Typical molecular weight cut-off (MWCO): roughly 1–500 kDa, with most drinking-water and wastewater UF membranes operating in the tens to low hundreds of kDa.
- Rejected: suspended solids, turbidity, bacteria, protozoa, algae, most colloids, membrane foulants, and high-molecular-weight organics. Virus retention varies with membrane characteristics and system integrity.
- Not rejected: dissolved salts, hardness ions (calcium and magnesium), dissolved orthophosphate, and short-chain dissolved polyphosphate.
“Dissolved” in this context usually means the fraction that passes through a 0.45 µm laboratory filter. That is a useful operational definition, because it separates what UF can physically screen out from what it cannot.
Polyphosphate Size Versus the UF Cut-Off
Polyphosphate is a family of condensed phosphate species, not a single molecule. Sodium hexametaphosphate, for example, is a mixture of chain and ring phosphates with varying chain length, and its average molecular weight depends on the grade. This matters, because retention is a size comparison, not a chemical reaction.
| Phosphorus species | Approximate size | Retention by a standard UF membrane |
|---|---|---|
| Orthophosphate (H₂PO₄⁻ / HPO₄²⁻) | Under 100 Da | None — passes freely |
| Short-chain polyphosphate (typical water-treatment grades) | Roughly several hundred to a few thousand Da | Little or none |
| Very long-chain polyphosphate | Above the membrane MWCO | Partial to substantial, depending on the membrane |
| Phosphate complexed with calcium, iron, or aluminium, or precipitated | Colloidal to particulate | Largely rejected once a solid phase forms |
| Phosphate adsorbed onto organic colloids or biomass | Colloidal | Partially rejected with the colloid |
The practical conclusion is simple: if your phosphate is genuinely dissolved and your membrane is a normal UF, plan for full passage. If your phosphate is particulate or colloid-bound, UF will remove it — but the removal is really particle removal, not phosphate removal.
Why Some Operators Still Measure Phosphate Removal
Reports of UF “removing” phosphate usually have one of four explanations. None of them turns UF into a phosphate-removal process.
1. Precipitation or complexation upstream of the membrane
If the feed water contains sufficient calcium, iron, or aluminium, phosphate can form sparingly soluble or colloidal species before it reaches the membrane. UF then removes those solids. The chemistry did the work; the membrane only filtered the result. Where no such reaction occurs, retention disappears.
2. Adsorption onto the membrane and the fouling layer
Fresh membranes and accumulated foulant layers can adsorb phosphate for a period. This produces a temporary removal signal that fades as sites saturate, and it can reverse sharply during chemical cleaning, when adsorbed phosphate is released back into the system. It is a batch effect, not a steady-state capability.
3. Concentration polarisation and sampling error
Phosphate accumulates near the membrane surface and is carried away in the concentrate stream. If feed, permeate, and concentrate are not all sampled at the same time under stable conditions, an apparent permeate reduction can appear that a proper mass balance does not support. Always check that the phosphate leaving in the permeate plus the concentrate accounts for the phosphate entering in the feed.
4. A tight membrane combined with long-chain polyphosphate
A low-MWCO UF membrane can retain the highest-molecular-weight fraction of a polyphosphate mixture while passing the shorter chains. The result is partial, selective retention that is difficult to predict and easy to misinterpret. It should be confirmed with membrane-specific test data, not assumed.
Where Polyphosphate Crystal Media Fits — and What It Does Not Do
Polyphosphate crystal media, sometimes referred to by trade names such as Siliphos, are slow-dissolving phosphate glasses used to release polyphosphate into a water line. Their function is scale inhibition and sequestration: the dissolved polyphosphate interferes with the growth of carbonate scale crystals and helps keep hardness ions dispersed in the water.
They are not softeners, and they are not filters. Hardness ions remain in the water. Phosphate dosing does not reduce total dissolved solids, does not remove iron or manganese, and does not disinfect. Those are separate treatment objectives that require separate equipment.
This has a direct consequence for UF system design:
- Polyphosphate released upstream of a UF membrane will largely pass through into the permeate, because it is dissolved.
- The media itself can shed fines, which UF will capture — increasing the load on the membrane and shortening the interval between backwashes.
- Dissolved polyphosphate gradually hydrolyses to orthophosphate, which is also dissolved and also passes UF. The phosphate does not disappear downstream; it simply changes form over time and residence length.
- Phosphate is a nutrient. Phosphate that passes through UF is available to biofilm in downstream RO membranes, storage tanks, and distribution piping. UF does not solve that problem, because it never removed the phosphate in the first place.
Where a polyphosphate medium and a UF membrane are both required, the sequence and the chemical compatibility should be evaluated as a system rather than as two independent components. Dosing location, hydrolysis time, and downstream membrane chemistry all interact.
If the Real Goal Is Phosphate Removal
Choose a process that targets dissolved ions, not one that targets particles:
- Nanofiltration or reverse osmosis: these reject dissolved ions, including phosphate, to a much greater degree than UF. Actual rejection depends on the membrane, pH, feed composition, and recovery.
- Chemical coagulation plus UF or MF: iron or aluminium salts convert dissolved phosphate into a particulate phase that the membrane can then reject. Coagulant dose and pH control are critical.
- Adsorption media: iron-oxide-based and alumina-based media can bind phosphate. Capacity, regeneration, and spent-media handling must be confirmed for the specific product.
- Anion exchange: effective on phosphate but sensitive to sulphate, chloride, and other competing anions.
- Biological phosphorus removal: appropriate for wastewater, not for a polishing loop on a clean water line.
Each option should be selected against the actual phosphorus speciation in the feed, because orthophosphate and polyphosphate do not behave identically in every process.
How to Verify Performance in Your Own System
- Sample feed, permeate, and concentrate at the same time under stable flow and temperature.
- Measure total phosphorus and dissolved phosphorus separately. Filter dissolved samples at 0.45 µm before analysis.
- Run a mass balance across the membrane skid. If the numbers do not close, the “removal” is a sampling artefact.
- Repeat the test after a chemical clean. A large phosphate release during cleaning indicates adsorption rather than rejection.
- Track the result over days, not hours. Adsorptive removal declines with time; size exclusion does not.
- Confirm the membrane MWCO and material with the membrane supplier, and confirm the polyphosphate grade and chain-length distribution with the chemical supplier.
Frequently Asked Questions
Does UF remove phosphate at all?
It removes particulate and colloid-bound phosphorus. It does not remove dissolved orthophosphate or dissolved short-chain polyphosphate.
Can a tighter UF membrane be used instead of RO for phosphate?
Not reliably. Even a low-MWCO UF membrane is a poor barrier to small dissolved anions. If dissolved phosphate removal is the objective, NF or RO is the appropriate membrane class.
Will polyphosphate crystal media protect downstream equipment without affecting a UF membrane?
The polyphosphate will pass through the UF rather than being removed by it, and the medium can contribute particulate fines. Whether the arrangement is acceptable depends on your sequence, water chemistry, and downstream equipment. Verify the specific configuration before committing to it.
Specifying the Right Solution
The distinction between scale inhibition, softening, filtration, and phosphate removal determines the equipment list. UF handles particles; polyphosphate crystal media handle scale tendency; neither removes dissolved phosphate. Getting that mapping right at the design stage avoids an expensive retrofit later.
If you are selecting polyphosphate crystal media, UF membranes, or a combined treatment train, ENVO CHEMICAL can review the application with you. To respond usefully, we need: the intended treatment objective, source-water analysis (hardness, alkalinity, iron, manganese, pH, and phosphorus speciation if available), the proposed process sequence and equipment type, membrane MWCO or media specification, design flow rate, and quantity required. Sharing these details up front keeps the discussion technical rather than speculative.
