Test for phase separation whenever something about the glass itself changes: its oxide composition, raw-material source, melting and cooling history, forming method, or post-treatment. Test as well when batch-to-batch dissolution behaviour, appearance, or field performance starts to drift. If none of those variables has moved and the process runs well inside a known single-phase region, routine phase-separation testing usually adds little. The difficulty is that polyphosphate glass development rarely stays still, and the changes that trigger amorphous phase separation are often the same ones made to improve dissolution rate, mechanical strength, or cost.

This article is written for R&D engineers, QA managers, and technical buyers working on polyphosphate water-treatment media — the slow-dissolving glass beads, crystals, or prills used for scale inhibition and corrosion control in potable and industrial water lines. It explains what phase separation is, why it matters for a dissolving medium, and exactly which development milestones should trigger a test.

What phase separation means in a polyphosphate glass

Amorphous phase separation is the splitting of a single, homogeneous glass into two (or more) glassy phases with different compositions. Both phases remain non-crystalline. It happens in two ways:

• Nucleation and growth — discrete droplets of one composition form inside a matrix of another.
• Spinodal decomposition — an interconnected, sponge-like structure develops with no distinct droplets.

Both can be completely invisible to the eye, or they can produce opalescence, haze, or a milky appearance if the domains grow large enough to scatter light.

What it is not

The term is often used loosely, so it is worth separating it from neighbouring phenomena:

Phenomenon	What it involves	How it is confirmed
Amorphous phase separation	Two glassy phases, no long-range order	Two glass transitions, no sharp XRD peaks, compositional contrast in electron imaging
Crystallisation (devitrification)	Ordered crystalline phases forming in the glass	Sharp peaks in XRD
Incomplete melting	Unreacted batch or refractory particles	Optical microscopy, EDS of the inclusions
Bubbles or striae	Gas voids or refractive-index streaks	Microscopy; often cosmetic rather than structural

It is also not a water-treatment function. Phase separation is a material property. Scale inhibition, corrosion inhibition, water softening by ion exchange, scale removal, and filtration are separate functions, and a polyphosphate medium does not perform all of them. Polyphosphate glasses sequester hardness ions and interfere with crystal growth, and they can contribute to protective film formation on metal surfaces; they do not remove calcium and magnesium from water the way a softening resin does, and they do not dissolve existing scale deposits. Phase separation matters because it changes how the glass releases its phosphate species over time.

Why phase separation is a practical problem for a dissolving medium

A polyphosphate medium is expected to dissolve predictably. Its whole value proposition rests on a controlled release rate over weeks or months, matched to the water volume, temperature, and hardness of a specific installation. Phase separation attacks that predictability in four ways:

• Uneven dissolution. The two phases have different solubilities. The more soluble phase leaches out first, leaving a depleted skeleton that releases far less phosphate than the early period did. The result is over-dosing at the start and under-dosing later.
• Batch-to-batch variability. A composition tolerance that is harmless in a single-phase glass can straddle an immiscibility boundary. Two production lots that meet the same chemical specification may then behave very differently in a filter housing.
• Mechanical breakdown. Internal phase boundaries and interconnected structures are weak paths. Beads and crystals can crack, crumble, or generate fines that migrate downstream and foul valves and fixtures.
• Invalid test data. If a dissolution or service-life study is run on a phase-separated sample, the resulting curve describes that sample — not the product you intend to manufacture at scale.

None of this means a phase-separated glass is automatically unusable. Some glasses are deliberately engineered to separate. For a dissolving water-treatment medium, the real questions are whether the release profile is acceptable and whether it can be reproduced lot after lot.

When to test: the trigger points

Treat each of the following as a reason to run at least a screening test. They are listed roughly in order of how often they cause trouble.

1. Any change to the oxide composition

Additions of alkaline-earth oxides, alumina, silica, boron, iron, or mixed alkalis are the classic route into an immiscibility region, because they distort or interrupt the phosphate chain network. Small substitutions count. A shift in the P2O5-to-Na2O ratio, a partial swap of one alkali for another, or a “minor” dopant added for strength or colour all qualify. Test at the laboratory melt stage, before the composition is frozen into a specification.

2. A change in raw-material source or grade

Impurities and hydration state change the effective composition and the melting behaviour. A new supplier, a different production lot, a different particle size of the same nominal chemical, or a switch between anhydrous and hydrated forms can all shift the melt. Screening is warranted even when the certificate of analysis looks equivalent.

3. Changes in melting and cooling

Phase separation depends on how long the melt or the cooling glass spends inside the immiscibility region, and at what temperature. Furnace residence time, melt atmosphere — water vapour in particular — crucible or tank material, quench medium, and cooling rate all influence the outcome. Faster quenching normally freezes a more homogeneous structure; slower cooling gives the structure time to separate. A change in any of these parameters is a trigger.

4. Scale-up and changes in forming method

Heat-transfer conditions differ enormously between a laboratory crucible, a pilot furnace, and a continuous production furnace. The same is true between casting into blocks, flaking, and droplet forming into beads or prills. Geometry matters as well: a thick block cools slowly at its core even when the surface sets quickly. Any transfer to a new line, new former, or new product size should be validated with a phase-separation check.

5. Post-processing and storage

Annealing, drying, coating, grinding, and packaging all involve thermal or mechanical history. A metastable single-phase glass can separate further during a heat step, and prolonged storage in hot, humid conditions can change the surface layer. If the product is dried or heat-treated after forming, test after that step, not only after quenching.

6. Field or QC signals

Sometimes the test is triggered by evidence rather than by a process change:

• Phosphate release that is high for the first days and then falls away sharply.
• Visibly hazy or milky media, or white deposits and fines inside cartridges.
• pH drift in treated water that cannot be explained by flow or hardness.
• Customer complaints about scale or corrosion control that survive a check of flow rate, water temperature, and hardness.
• Dissolution results that wander outside the normal control window between otherwise identical lots.

Change made	Relative risk	Minimum useful test
Dopant added for strength or colour	High	Thermal analysis plus dissolution curve against reference
New raw-material supplier	Medium to high	Appearance, density, dissolution curve
Longer furnace residence time	High	Thermal analysis, electron imaging
Quench rate altered	High	Thermal analysis, dissolution curve
New forming line or product size	Medium	Cross-section imaging, dissolution curve
Packaging or storage change only	Low	Appearance and retained-sample comparison

A practical test sequence

Phase separation is rarely proven by a single instrument. Build the evidence in tiers so that most decisions are made with inexpensive methods.

1. Screening. Visual inspection under strong oblique light, optical microscopy with an immersion liquid, density, and a short dissolution run compared against a retained reference lot. This tier catches gross problems and should be part of routine QC anyway.
2. Confirmation. Thermal analysis (DSC or TMA) to see whether one glass transition or two are present; XRD to rule out crystallisation; scanning electron microscopy in backscattered mode with EDS mapping to reveal compositional contrast across a polished cross-section.
3. Characterisation. For features below roughly a hundred nanometres, transmission electron microscopy with EDS, small-angle X-ray or neutron scattering to quantify composition fluctuations, and solid-state 31P MAS NMR to reveal changes in the phosphate species distribution and connectivity. Longer-duration leaching or durability testing closes the loop.

Reading the results honestly

Two glass transitions with no sharp XRD peaks and clear compositional contrast together point to amorphous phase separation. Sharp XRD peaks mean crystallisation instead, which is a different defect with different corrective actions. Opalescence on its own proves nothing — it can come from bubbles, unmelted batch, or fine crystallites. And a positive phase-separation result does not by itself condemn the material. What matters is whether the dissolution profile still meets the application requirement and whether it repeats.

For any specific product, the parameters that must be confirmed rather than assumed include the actual oxide composition and its tolerance band, the melting and cooling route, the target release rate and service life under defined water conditions, and — where drinking-water contact is intended — suitability for that exact product, application, and destination market. A food-grade designation does not by itself establish drinking-water suitability.

When testing is probably unnecessary

Skip the additional work when nothing in the trigger list has changed: same composition, same suppliers, same melting and cooling practice, same forming route, same post-treatment. If the composition sits comfortably inside a single-phase region that has already been mapped, and routine QC includes an appearance check and a dissolution test sensitive enough to catch drift, added phase-separation testing usually duplicates existing evidence. The moment any of those constants moves, the calculation changes.

Keep a reference, not just a result

The most useful phase-separation data is comparative. Retain a sample from each production lot, record composition and tolerance, raw-material lot, melting and cooling parameters, forming method, and the reference dissolution curve. A new lot can then be judged against a known-good baseline rather than against an absolute threshold that may not exist for your glass system. This is also the fastest way to answer a field complaint: compare the suspect material with the retained sample from the same lot and with the qualification lot.

Frequently asked questions

Is phase separation the same as crystallisation?

No. Phase separation produces two glassy, non-crystalline regions; crystallisation produces ordered phases. The tests differ — XRD distinguishes them quickly — and the corrective actions are not the same.

Can I detect phase separation by looking at the media?

Sometimes. Droplet-type separation with large domains scatters light and looks hazy. Fine spinodal structures are invisible, and a clear bead can still be phase-separated. Appearance is a trigger for testing, not a substitute for it.

Does a phase-separated polyphosphate medium always fail?

Not automatically. It usually hurts reproducibility of the release rate and can weaken the media, but whether that is acceptable depends on your dissolution specification and the installation. The decision should be based on release data and mechanical integrity, not on the label alone.

My supplier calls the product “siliphos crystal.” Does that tell me the composition?

Not reliably. The term is used loosely in the trade, and products sold under it or similar names vary in composition, form, and dissolution behaviour between suppliers. Specifications and test data must come from the specific supplier for the specific product, and terminology should never be treated as a substitute for a data sheet.

Planning your next development step?

If you are qualifying a polyphosphate glass medium and need to know whether a composition or process change is likely to push the material into a two-phase structure, tell us the essentials: your intended application and target release behaviour, current or proposed oxide composition and tolerance, melting and cooling route (including quench method), the physical form and size you need, your source-water conditions where relevant, anticipated volumes, and the destination market. With that information, ENVO CHEMICAL can discuss the appropriate material specification and the testing evidence that should accompany it.