They Call It All “Flux” — So Why Does One Brighten the Glaze and Another Turn It Matte?

 Nov 15, 2026

 Leave a message

 

Peak temperature decides how it melts. Cooling decides what it becomes.

When I first started formulating glazes, I treated "fluxing" as a single job: lower the melting point so the glaze spreads out at temperature.

It didn't take long to learn that this wasn't true.

Swap one feldspar for another and the glaze suddenly goes clear and glassy. Nudge the calcium or the zinc and the gloss drops away, leaving a soft haze. What really didn't make sense was this: same formula, same peak temperature, only a slightly slower cooling ramp - and the glaze came out matte.

Back then I blamed the raw materials, or the addition levels. I changed suppliers, recalculated the batch, and got nowhere.

What I eventually came to understand is that a flux only governs melting at high temperature. Whether the glaze ends up bright or matte is settled during cooling. Those are two separate problems.

 


1. Start by defining gloss

 

Without a definition of gloss, everyone is talking past each other.

Gloss is a glaze surface's ability to reflect light specularly. It depends on exactly two things: how flat the surface is, and how uniform the glaze layer is throughout.

Once surface roughness reaches the scale of visible-light wavelengths, the reflected light is scattered and the eye reads the surface as matte. If crystalline phases, phase-separated droplets or fine bubbles are present inside the layer, light enters and doesn't come back out - the result looks dull.

So when a glaze goes matte, there are only three directions to investigate: the surface never leveled, crystals formed within the layer, or something is scattering light inside that isn't crystalline. Every raw material and every firing curve ultimately comes back to one of those three.

A note on measurement. A gloss figure means nothing without the angle. For ceramic tiles the industry convention is 60°. On the same tile, readings at 20° and 85° can differ by more than twenty points, so an unqualified "gloss value" isn't comparable to anything. Rough bands at 60°: above 70 is high gloss, 30–70 semi-gloss, 10–30 satin, below 10 matte. Standards vary from company to company; those are the orders of magnitude.

High whiteness wollastonite filler

 


2. The same oxide plays two roles

 

In the melt, K₂O, Na₂O, CaO, MgO and ZnO all do the same job: they break up the silicate network, drop the viscosity and let the glaze flow out. There, calling them fluxes is correct.

But they are still in the glaze when it cools. Calcium combines with alumina and silica to form anorthite; zinc combines with silica to form willemite.

The role has changed - from network breaker to crystallizing component.

The same oxides flux at temperature and crystallize on cooling. What separates the two roles is temperature - and time.

Which means the question "is this material a brightening flux or a matte flux?" is badly posed. There is no fixed answer, only a tendency under a given firing schedule.


 

3. Potassium and sodium: reliable brighteners, and cooling isn't the problem

 

K₂O and Na₂O are the classic brightening fluxes. They cut viscosity sharply at temperature, the glaze levels out well and forms a continuous, fully melted glass. If you want high gloss, it's hard to avoid potassium and sodium.

The glass they form is stable on cooling. Under normal slow cooling it doesn't devitrify, and the gloss holds.

The trouble lies elsewhere.

First, thermal expansion. Potassium and sodium push the expansion coefficient up, and once the body–glaze mismatch is large enough you get crazing and crackle. Crazing, note - not dunting. Dunting is the body cracking apart during cooling. A different failure entirely.

Second, water resistance. When total alkali is high and there isn't enough Al₂O₃ to lock it into the network, water resistance drops and the surface can bloom white over time. That isn't "matte," but it looks just as bad.

 

Third, viscosity. Overdo it and the melt becomes too fluid - the glaze runs and builds up at edges.

 

One point that is often missed: don't use K and Na singly. Blending them (the mixed-alkali effect) gives lower thermal expansion and better water resistance than either alone. At the same 4% total alkali, all-sodium versus a 2:1 K:Na blend will differ noticeably in crazing rate.

On the cooling side, the only way potassium and sodium cause trouble is if cooling is too fast. Stress in the layer can't relax, fine microcracks open up, and the surface looks hazy from a distance. That isn't crystallization - don't go looking for a formula fix. Slow the cooling ramp, or bring the expansion down, and it comes back. Sodium is more reactive than potassium, and raw-material batch variation shows up more clearly through the sodium.


 

4. Calcium and magnesium: the "threshold" rule of thumb has led plenty of people astray

 

For calcium to turn a glaze matte, three conditions have to be true at once.

First, there must be enough Al₂O₃. The primary crystal calcium forms is anorthite (CaAl₂Si₂O₈), and anorthite consumes alumina. Without enough alumina, no amount of calcium will grow anorthite - it stays in the glass and keeps fluxing.

Second, there must not be enough alkali to suppress crystallization. Enough K and Na, and the tendency to crystallize is pushed down.

Third, the glaze has to move slowly enough through the crystallization window. Anorthite crystallizes roughly between 950 and 1100°C, and that window shifts with the formula. Slow down through it and the crystals have time to grow.

Miss any one of the three and calcium gives you a bright glaze back.

So to judge whether a calcium system will go matte, don't look at the CaO percentage. Look at the CaO:Al₂O₃ ratio, and look at the cooling curve. You can make a bright glaze at 12% CaO, and a matte one at 7% - the difference is all in the alumina and the cooling.

Magnesium follows a different curve. MgO raises high-temperature viscosity and widens the firing range; the phases it crystallizes are diopside (CaMgSi₂O₆) or enstatite, which need higher temperatures and longer holds to appear. Which is why "magnesium's threshold is a few points lower than calcium's" is a meaningless rule of thumb - they don't operate in the same temperature window at all.

The shop-floor version is: calcium decides whether it goes matte, magnesium decides how evenly. Add talc and the matte becomes finer and softer, but whether it goes matte at all still comes down to calcium and cooling.


 

5. Zinc: the most contradictory - and the most cooling-dependent

 

Zinc is the most contradictory of the group.

In small amounts it lowers viscosity and improves wetting on the body; the glaze comes out warm and luminous, and it is an excellent brightening component.

Past a certain level it becomes the main crystallizing species, precipitating willemite (Zn₂SiO₄) and producing a crystalline matte outright.

That "certain level" has no universal number - it is tied to the alumina-silica ratio, the alkali content and the cooling. The same zinc content on a different kiln gives a different result, which is exactly what makes zinc-bearing frits so troublesome: the supplier's certificate of analysis is perfectly clear, but it was written for one kiln line and doesn't necessarily hold on yours.

Willemite crystallizes roughly between 900 and 1050°C, and the window shifts with zinc content and alumina-silica ratio. The cooling rate through that window is essentially the gloss switch for a zinc glaze. At the same zinc level, 10°C/min comes out bright and 2°C/min comes out matte.

The reliable approach is to split that interval into several rates and run trials rather than guess. Fix the formula, the application weight and the peak temperature; change only the 1050→900°C cooling rate - say 10, 6 and 3°C/min - and read each tile on a 60° gloss meter. Put the three side by side and you have located the switch for zinc in that system.

One more thing that is easy to miss: when alumina is high, zinc may not form willemite at all - it may form gahnite (ZnAl₂O₄). Both look matte, but the crystal is different and so is the fix: one is a zinc adjustment, the other an alumina adjustment. Reducing zinc simply because "there's too much zinc" sometimes won't bring the gloss back, because the problem was the alumina.

Zinc systems have another trap on fast-fire lines: zinc tends to induce pinholes and blisters. That is a separate issue from crystallization - one happens on the way up, the other on the way down. Don't try to fix them together.


 

6. Boron: it's not "too much boron," it's "not enough alumina"

 

Boron is a special case. It is both a network former and a flux: it gives low melt viscosity, good wetting and easy bubble release, which helps produce a clean, transparent surface.

But boron behaves counterintuitively. At low levels it enters the network as [BO₄] tetrahedra, which actually raises viscosity and lowers expansion; only past a certain point do [BO₃] triangles take over, break the network and bring viscosity down. "More boron means a thinner melt" is true only in the second regime.

-5

There are two main pitfalls.

The first is bubble release. Boron sources differ enormously. Borax decahydrate carries only about 18% B₂O₃ in theory - the rest is water of crystallization; boric acid carries 56% and loses more than 40% on firing. If you are formulating for 5% B₂O₃, borax decahydrate means adding something like 27 parts - and all that water has to escape during heating. If it can't, you get pinholes. Those 27 parts also bring in more than 4 parts of Na₂O, so the alkali in the formula shifts too. Boron frit is far cleaner: it has already been melted and quenched, so loss on ignition is low.

The second is phase separation. High boron, low alumina and a little too much alkali, combined with slow cooling, will separate the glaze into a silica-rich phase and a boria-rich phase. When the droplet size lands near visible wavelengths, scattering peaks and the glaze looks grey and flat - an opacifying haze. It still feels smooth to the touch, completely unlike a crystalline matte.

But don't believe "above 5% boron it phase-separates." Bright glazes with more than 5% B₂O₃ are extremely common in fast-fire wall and floor tile; as long as the alumina keeps up, phase separation stays suppressed.

 

 

Verifying phase separation is simple: fire two pieces from the same batch, one quenched and one slow-cooled. If the slow-cooled piece looks cloudy and the quenched one is clear, it is phase separation, not crystallization.

 

Oxide behavior and the tendencies of each flux family only help you predict a direction. What actually decides whether a given glaze comes out bright or matte is the combination of levels, the alumina-silica ratio, the raw material form, and that invisible curve inside the kiln.

Paper analysis can save you a few firings. It can't replace the test tile.

 

If you have run into a case where the formula barely changed and only the cooling was adjusted, yet the glaze came out completely different, send over the base formula, the raw material analysis, the firing curve and the test readings - let's look at it together.

Send Inquiry
Send Inquiry