Ceramic glazes melt and flow according to their chemistry and mineralogy. Observing and measuring the nature and amount of flow is important in understanding them.
Glazes become fluid when they melt, they are molten. The fluidity (or viscosity) of this melt needs to be considered, especially when troubleshooting problems. While two different fired glazes may appear to have melted a similar amount (even on a vertical surface), one may be radically more fluid than the other (this becomes evident in a fluidity tester or when the glaze is applied thicker). While it might seem logical that a matte glaze has a fairly stiff (viscous) melt, it might actually be highly fluid and runny (because the matteness is usually a product of crystallization on the surface during cooling or phase separation within the glass). Melt flow testers are an ideal way to get a true picture of how fluid a glaze melt really is. In a well-designed melt flow tester a glaze with the correct degree of melt flow will travel half way down the runway. Testing melt fluidity using balls and cone shapes also provides valuable information.
Glazes do not simply melt (unless they are 100% frit). Certain particles in the matrix do melt. As they become more and more active they dissolve the refractory particles around them (e.g. quartz/silica, clay). That means that the size of the refractory particles and the time available determine the completeness of the melting. Potters normally fire slow enough that it is assumed that all particles have gone into solution if a glaze appears well melted. But in industry, where firings go from cold to cold in two hours, attention must be given to this. For this reason the use of frits and ball milling to reduce silica particle sizes are standard practice.
Glaze melt fluidity relates closely to a variety of problems like pinholing, crawling, gloss, blistering, crazing and even leaching. Logically, glazes for vertical surfaces will be more viscous than tile glazes, for example, which are applied to horizontal surfaces. Molten glaze viscosity can be understood in terms of molecular silicate chains (which also link across to other chains). The chemistry of the melt (and the degree to which materials have released their oxides to it) determines the rigidity of the structure and therefore the viscosity of the melt. Glazes high in powerful fluxes (like boron, lithium, sodium) melt and run more. In functional ware, for example, it is desirable to have enough melt to bring into solution all the material particles and produce a fired surface that has good gloss. However if too much flux is present the fired glaze is not as hard, it can have higher thermal expansion (if it contains high KNaO), may be more prone to blistering and is more likely to leach. Thus it is best to tune the ratio of fluxes to SiO2 and Al2O3 such that the melt has the right degree of movement and no more. Even special purpose reactive or matte glazes need to be tuned. In the case of the former, a compromise is needed between the high fluidity needed to produce the visual effect and a more stable and harder stiffer melt. For matte glazes, a less fluid type that relies more on high MgO rather than high Al2O3 only will have less cutlery marking of the fired glass.
Blistering often occurs in glazes of high melt fluidity. This might appear illogical since it would seem that such melts would more readily pass gases of decomposition from the body. However, the problem often happens because these glazes begin to melt (and seal the body surface) at much lower temperatures than one might think. Then they just keep percolating the escaping gases as the kiln is soaked and even continue after the kiln is shut off. Fast dropping temperatures finally freeze these blisters into the glass at an even lower temperature than they first melted at. Employing less gassing materials, applying a denser laydown of glaze, a flux system that melts later or firing the kiln down somewhat and then soaking might be the solution.
The Potter's dictionary has a very good discussion with diagrams of this under the term 'viscosity'.
10 grams GBMF test balls of these three glazes were fired to cone 6 on porcelain tiles. Notice the difference in the degree of melt? Why? You could just say glaze 2 has more frit and feldspar. But we can dig deeper. Compare the yellow and blue numbers: Glaze 2 and 3 have much more B2O3 (boron, the key flux for cone 6 glazes) and lower SiO2 (silica, it is refractory). That is a better explanation for the much greater melting. But notice that glaze 2 and 3 have the same chemistry, but 3 is melting more? Why? Because of the mineralogy of Gerstley Borate. It yields its boron earlier in the firing, getting the melting started sooner. Notice it also stains the glaze amber, it is not as pure as the frit. Notice the calculated thermal expansion: That greater melting came at a cost, the thermal expansion is alot higher so 2 and 3 glaze will be more likely to craze than G2926B (number 1).
Albany Slip was a pure mined material, Alberta Slip is a recipe of mined materials and refined minerals designed to have the same chemistry, firing behavior and raw physical appearance.
These melted-down-ten-gram GBMF test balls of glaze demonstrate the different ways in which tiny bubbles disrupt transparent glazes. These bubbles are generated during firing as particles in the body and glaze decompose. This test is a good way to compare bubble sizes and populations, they are a product of melt viscosity and surface tension. The glaze on the top left is the clearest but has the largest bubbles, these are the type that are most likely to leave surface defects (you can see dimples). At the same time its lack of micro-bubbles will make it the most transparent in thinner layers. The one on the bottom right has so many tiny bubbles that it has turned white. Even though it is not flowing as much it will have less surface defects. The one on the top right has both large bubbles and tinier ones but no clouds of micro-bubbles.
Fired to cone 10 oxidation. Although feldspar is a key melter in high and medium temperature glazes, by itself it does not melt as much as one might expect in this GLFL test. The Montana materials on the right are not commercially available, they were being evaluated for viability.
A example of a highly fluid cone 6 glaze that has pooled in the bottom of a mug (and crystallized). Glazes normally need to be under some compression to avoid crazing (by having a lower-than-the-body thermal expansion), but if they are thick like this the body does not have the strength to resist the extra outward pressure the glaze can be exerting at the base from the inside. The result here is a separated base. Conversely, if the glaze is under tension (having too high an expansion), the cracks that develop within it to relieve the tension are deep and wider and thus more likely to propagate into the body below. The ultimate result: Poor ware strength.
An example of a highly fluid glaze melt that has pooled in the bottom of a bowl. The fluidity is partly a product of high KNaO, thus it is also crazed (because KNaO has a very high thermal expansion). While it may to decorative, this effect comes at a cost. The crazing weakens the piece, much more than you might think (200%+). Those cracks in that thick layer at the bottom are deep, they want to continue down into the body and will do so at the first opportunity (e.g. sudden temperature change, bump). Also, fluid glazes like these are more likely to leach.
In the glaze on the left (90% Ravenscrag Slip and 10% iron oxide) the iron is saturating the melt crystallizing out during cooling. GR10-K1, on the right, is the same glaze but with 5% added calcium carbonate. This addition is enough to keep most of the iron in solution through cooling, so it contributes to the super-gloss deep tenmoku effect instead of precipitating out.
The glaze is cutlery marking (therefore lacking hardness). Why? Notice how severely it runs on a flow tester (even melting out holes in a firebrick). Yet it does not run on the cups when fired at the same temperature (cone 10)! Glazes run like this when they lack Al2O3 (and SiO2). The SiO2 is the glass builder and the Al2O3 gives the melt body and stability. More important, Al2O3 imparts hardness and durability to the fired glass. No wonder it is cutlery marking. Will it also leach? Very likely. That is why adequate silica is very important, it makes up more than 60% of most glazes. SiO2 is the key glass builder and it forms networks with all the other oxides.
This cone 10R glaze, a tenmoku with about 12% iron oxide, demonstrates how iron turns to a flux in reduction firing and produces a glaze melt that is much more fluid. In oxidation, iron is refractory and does not melt well (this glaze would be completely stable on the ware in an oxidation firing at the same temperature, and much lighter in color).
Plainsman M340 Transparent liner with various stains added (cone 6). These bubbles were fired on a bed of alumina powder, so they flattened more freely according to melt flow. You can see which stains flux the glaze more by which bubbles have flattened more. The deep blue and browns have flowed the most, the manganese alumina pink the least. This knowledge could be applied when mixing these glazes, compensating the degree of melt of the host accordingly.
These are the same glaze, same thickness, Ulexite-based G2931B glaze, fired to cone 03 on a terra cotta body. The one on the right was fired from 1850F to 1950F at 100F/hr, then soaked 15 minutes and shut off. The problem is surface tension. Like soapy water, when this glaze reaches cone 03 the melt is quite fluid. Since there is decomposition happening within the body, gases being generated vent out through surface pores and blow bubbles. I could soak at cone 03 as long as I wanted and the bubbles would just sit there. The one on the left was fired to 100F below cone 03, soaked half an hour (to clear micro-bubble clouds), then at 108F/hr to cone 03 and soaked 30 minutes, then control-cooled at 108F/hr to 1500F. During this cool, at some point well below cone 03, the increasing viscosity of the melt becomes sufficient to overcome the surface tension and break the bubbles. If that point is not traversed too quickly, the glaze has a chance to smooth out (using whatever remaining fluidity the melt has). Ideally I should identify exactly where that is and soak there for a while.
Wrong. It is the one on the right. While the copper looks so much better in that fluid one on the left, that melt mobility comes at a cost: blisters. As a clear glaze it is no glossier than the other one, but it runs into thicker zones at the bottom and they blister. This is because the high mobility coupled with the surface tension blows bubbles as gases of decomposition travel through (in a normal cooling kiln they break low enough that mobility is insufficient to heal them). The fired glass in the one on the left is also not as hard, it will be more leachable, it will also craze more easily and be more susceptible to boron-blue devritrification. But as a green? Yes it is better.
I am comparing 6 well known cone 6 fluid melt base glazes and have found some surprising things. The top row are 10 gram GBMF test balls of each melted down onto a tile to demonstrate melt fluidity and bubble populations. Second, third, fourth rows show them on porcelain, buff, brown stonewares. The first column is a typical cone 6 boron-fluxed clear. The others add strontium, lithium and zinc or super-size the boron. They have more glassy smooth surfaces, less bubbles and would should give brilliant colors and reactive visual effects. The cost? They settle, crack, dust, gel, run during firing, craze or risk leaching. In the end I will pick one or two, fix the issues and provide instructions.
The top samples are 10 gram GBMF test balls melted down onto porcelain tiles at cone 6 (this is a high melt fluidity glaze). These balls demonstrate melt mobility and susceptibility to bubbling but also color (notice how washed out the color is for thin layers on the bottom two tiles). Both have the same chemistry but recipe 2 has been altered to improve slurry properties. Left: Original recipe with high feldspar, low clay (poor suspending) using 1.75% copper carbonate. Right: New recipe with low feldspar, higher clay (good suspending) using 1% copper oxide. The copper oxide recipe is not bubbling any less even though copper oxide does not gas. The bubbles must be coming from the kaolin.
This is an example of how useful a flow tester can be to check new glaze recipes before putting them on ware and into your kiln. This was fired to only cone 4, yet that fritted glaze on the left is completely over-melted. The other one is not doing anything at all. These balls are easy to make, you only need weigh out a 50 gram batch of glaze, screen it, then pour it on a plaster bat until it is dewatered enough to be plastic enough to roll these 10 gram balls.
A GLFL test for melt-flow to compare Custer Feldspar from Feb/2012 (right) with Mar/2011 (fired at cone 6). Custer Feldspar does not melt like this by itself at cone 10. It was mixed 80:20 Feldspar:Ferro Frit 3134. This test demonstrates that the material has been very consistent between these two shipments.
This is G2926B cone 6 transparent glaze. I am developing a simple test procedure to produce an absolute measurable value for glaze melt flow and it appeared it would be worthwhile to create a mold to make these cone-shaped samples. But I was wrong. Both specimens are exactly 10 grams, but the simple ball flows better. This is likely because of better early heat penetration because there is only a small area of contact with the tile.
Here it is fired to cone 8 where the melt obviously has much more fluidity! The photo does not do justice to the variegation and crystallization happening on this surface. Of course it is running alot more, so caution will be needed.
This is Linda Arbuckle's base recipe (66% frit 3124, 23% feldspar, 13 kaolin/bentonite, 9 zircon, 4 tin oxide mixed to 1.62 specific gravity). It is fired at cone 05 creating a super gloss. This is applied very, very thickly (double the thickness of what a stoneware glaze would be). Yet notice how the air-vent holes did not heal during firing, these would have filled in easily had this been on stoneware or porcelain. For this reason, majolica glazes must be applied to ware not having any abrupt concave contours (this also happened on the handle-joins and where foot-meets-wall inside). Bisque must be clean to assure good adherence. Any glaze coverage issues must be repaired carefully before firing. This piece was bisque fired at cone 06, that is the reason for the air vent holes. Had it been bisque fired at cone 04 (higher than the glaze firing) these would not have occurred. However that would have extended the dip-time, this one was held under 10 seconds whereas a cone 04 bisque would have required 20 seconds.
Why are these happening (on this piece by Paul Briggs)? It is not completely clear. The glaze has plenty of carbonates, including copper, enough for over 20% LOI. But these normally produce high populations of small blisters, this is the opposite. The melt appears to have enough surface tension that the bubbles survive and endure top-temperature-soaking. And they don't pop until the temperature has dropped so far that insufficient melt-mobility remains to heal them. The glaze has an unconventionally low SiO2 content, that makes it flow vigorously, well enough that the melt is moving and collecting in surface contours. The glaze recipe is quite unconventional, any effort to "improve" its adherence to limits would likely lose the visual aesthetic. A drop-and-hold firing schedule is likely the key to alleviating this.
The height down to which the cone melts is measured and recorded. Courtesy of Ashok Srivastava.
This recipe melts to such a fluid glass because of its high sodium and lithium content coupled with low silica levels. Reactive glazes like this produce interesting visuals but these come at a cost that is more than just the difficulty in firing. Recipes like this often calculate to an extremely high thermal expansion. That means that not only will this form a lake in the bottom of ware when used on the inside, but the food surfaces will craze badly. The low silica will also contribute to leaching of the lithium and any colorants present.
The boron and zinc fluxes make the melt of this glaze highly fluid at cone 7R. That comes with consequences. Notice the Al2O3 and SiO2 in the calculated chemistry. They are at cone 04 levels. The significant ZnO increases surface tension of the melt, this helps bubbles form at the surface (like soap in water). Al2O3 and SiO2 could be added (via more clay), this would stiffen the melt so the large bubbles would be less likely to form (this glaze melts so well that it could accept significantly more clay without loss in gloss). A drop-and-soak firing is another option, in this case a drop of more than 100C might be needed (see the link below to learn more).
This glaze creates the opaque-with-clear effect shown (at cone 7R) because it has a highly fluid melt that thins it on contours. It is over fired. On purpose. That comes with consequences. Look at the recipe, it has no clay at all! Clay supplies Al2O3 to glaze melts, it stabilizes it against running off the ware (this glaze is sourcing some Al2O3 from the feldspar, but not enough). That is why 99% of studio glazes contain clay (both to suspend the slurry and stabilize the melt). Clay could likely be added to this to increase the Al2O3 enough so the blisters would be less likely (it would be at the cost of some aesthetics, but likely a compromise is possible). There is another solution: A drop-and-soak firing. See the link below to learn more. One more observation: Look how high the LOI is. Couple that with the high boron, which melts it early, and you have a fluid glaze melt resembling an Aero chocolate bar!
Fired at cone 6. A melt fluidity comparison (behind) shows the G3808A clear base is much more fluid. While G2926B is a very good crystal clear transparent by itself (and with some colorants), with 2% added copper oxide it is unable to heal all the surface defects (caused by the escaping gases as the copper decomposes). The G3808A, by itself, is too fluid (to the point it will run down off the ware onto the shelf during firing). But that fluidity is needed to develop the copper blue effect (actually, this one is a little more fluid that it needs to be). Because copper blue and green glazes need fluid bases, strategies are needed to avoid them running off the ware. That normally involves thinner application, use on more horizontal surfaces or away from the lower parts of verticals.
Each potter using Tenmoku has their own preferences about how the glaze should look. Ron clearly likes the iron crystals to develop well on the edges of contours. He has learned how to walk a delicate firing and recipe balance to achieve this effect. If the percentage of iron is too high, or the glaze is applied too thin, reduction is too heavy or the cooling too slow there will be too muchy crystallization. If the iron is too low, cooling is too fast or the glaze it too thick it will be a solid black. Additionally, this effect depends on a glaze having a fluid melt (the iron is a strong flux), if the glaze is too thick it will run downward during the firing.
These two glazes are both brilliant glass-like super-transparents. But on this high-iron stoneware only one is working. Why? G3806C (on the outside of the piece on the left) melts more, it is fluid and much more runny. This melt fluidity gives it the capacity to pass the micro-bubbles generated by the body during firing. G2926B (right) works great on porcelain but it cannot clear the clouds of micro-bubbles coming out of this body. Even the glassy smooth surface has been affected. The moral: You need two base transparents in which to put your colors, opacifiers and variegators. Reactive glazes need melt fluidity to develop those interesting surfaces. But they are more tricky to use and do not fire as durable.
These Plainsman Midstone and Redstone cups are fired to cone 6 with M340 Transparent glaze liner (these are raw materials that body manufacturers incorporate into their products in fairly high percentages). Notice how many more glaze bubbles there are with the red cup. This is typical using other transparent glazes also. To get a bubble-free clear on this red burning body a glaze having a higher melt fluidity is needed.
Both mugs have the same cone 6 oxidation high-iron (9%), high-boron, glossy glaze. Iron silicate crystals have completely invaded the surface of the one on the right, turning the near-black glossy into a yellowy matte. Why? Three things. It was slow-cooled and the other free-fall-cooled (firings done in the same kiln). The glaze has a fluid melt (it runs) and its percentage of iron is high enough that it could precipitate out from solution in the melt (given the time). Susceptible glazes have a temperature at which crystals form the best and that temperature can be hundreds of degrees down from the firing cone (or higher if precipitation is occurring). In industry, devitrification is regarded as a defect. But potters call it crystallization. Understanding (especially the chemistry and materials) and experimental firings are needed to learn to control and exploit the effect in a glaze.
The glaze is running down on the inside, so it has a high melt fluidity. "High melt fluidity" is another way of saying that it is being over fired to get the visual effect. It is percolating at top temperature (during the temperature-hold period), forming bubbles. There is enough surface tension to maintain them all the way down to the body, and for as long as the temperature is held. To break the bubbles and heal up after them the kiln needs to be cooled to a point where decreasing melt fluidity can overcome the surface tension. The hold temperature needs to be high enough that the glaze is still fluid enough to run in and and heal the residual craters. A typical drop temperature is 100F.
These glazes are the same (G3806G), except the one on the right has 3.5% copper carbonate added. Copper is commonly thought to flux glazes, making them melt more. But in this case, the clear base is running just as much as the stained one. And I was suspicious that the micro-bubbles in the glass matrix were coming from the copper carbonate gassing during firing. But not so, as you can see the flow on the left has them also, actually it has even more.
This is G3806F copper green glaze. It is highly melt fluid at cone 6. As the glaze begins to melt notice that it splits on the sharp concave sections on the handle join. The underscore the needs for the melt to be sufficiently fluid to flow back in and heal these areas.
Both pieces are the same clay, same glaze. The one on the left went to cone 4. Notice how full of holes and bubbles the glaze is. The one on the right went to cone 6 using the C6DHSC firing schedule. It is perfectly smooth and glassy.
This is recommended in the booklet "15 Tried and True Cone 6 Glaze Recipes". This melt flow tester compares it with a typical cone 6 glossy, G2926B. This recipe is 90% Frit 3110 and 10% kaolin and their booklet recommends adding stains to it. But anyone knowing a little about this frit knows it would run off this flow tester even before bisque temperatures. It is crazy to recommend this. Even as a crackle. For cone 6 it needs to be diluted much more, not just with kaolin but also silica. I knew this would run but I underestimated its melt fluidity. I put a large tile below the tester to catch overrun, yet the melt ran off that and a big three-cm-wide blob melted through the kiln wash and so far into my zircon shelf I cannot chip it off! I cannot imagine how many people have tried this on vertical surfaces and had the same thing happen. The lesson: Use common sense when looking at recipes, then you don't even need waste time testing them. Even if their authors did not!
A cone 6 firing. The glaze on the left has a B2O3 molar content of 0.54 whereas the one on the right has 0.64 (other oxide levels are the same). This is triple the typical amount of boron in a cone 6 glaze, the result is obvious: High melt fluidity for both. But G3904A has a significant characteristic that is different: The flow is more transparent because of the lower micro-bubble population. It's melt is less viscous, that enables the bubbles to pass, exit and the surface to heal. Why don't all glazes use more boron? Cost. Frits are expensive and they are the best source of boron. There is also a cost to durability (although mitigated when there is plenty of Al2O3 and SiO2 present, as is the case here). These recipes were part of an interesting project to fix a recipe where the potter mistakenly used Frit 3134 instead of 3124 when mixing a large batch of glaze. I calculated how much kaolin and silica to add to bring the chemistry back into line with the original. This was possible because frit 3134 chemistry is an approximate oxide-subset of 3124. The resultant glaze is potentially better than the original.
These two Plainsman M370 mugs were fired at cone 6, the left one with G2934 matte glaze, the right one with G2934Y4 matte. They look and feel identical in the hand. The two glazes have the same chemistry. But they employ different materials to source that chemistry. The secret of of the matteness is high MgO (magnesia content). In the glaze on the left that MgO is sourced by dolomite, a lot of it. The glaze on the right sources it from a special frit, Ferro 3249. The impact of this difference is visible in the melt flows, the fritted one is obviously melting and flowing better. On other clays, especially stonewares, the G2934 can have a dry surface that cutlery marks. Thicker applications make it worse. But the Y version exhibits no such issues. Its mattness, durability, cleanability and hardness are so good that it is being used in floor tile.
I am going to fire them at about 10 temperatures starting at cone 022. These will be a great to determine whether they suddenly melt or slowly soften over a range of temperatures. I will be able to see how wide that range is and how they behave as they go beyond it. I also included Gerstley Borate, mother nature's nature boron source. They way it melts should clear contrast with the frits.
Fired at 1850. Notice that Frit 3195 is melting earlier. By 1950F, they appear much more similar. Melting earlier can be a disadvantage, it means that gases still escaping as materials in the body and glaze decompose get trapped in the glass matrix. But if the glaze melts later, these have more time to burn away. Glazes that have a lower B2O3 content will melt later, frit 3195 has 23% while Frit 3124 only has 14%).
In ceramics, glazes melt to produce a liquid glass. That glass exhibits surface tension and it is important to understand the consequences of that.
In ceramics, glazes and bodies have a chemistry, a mineralogy and a physical presence. All of these need to be understood to adjust and fix issues.
Fluid Melt Glazes
Fluid melt glazes and over-melting, over fired, to the point that they run down off ware. This feature enables the development of super-floss and cyrstallization.
The melting temperature of ceramic glazes is a product of many complex factors. The manner of melting can be a slow softening or a sudden liquifying.
In ceramic slurries (especially casting slips, but also glazes) the degree of fluidity of the suspension is important to its performance.
Questions and suggestions to help you reason out the real cause of ceramic glaze blistering and bubbling problems and work out a solution
A Low Cost Tester of Glaze Melt Fluidity
This device to measure glaze melt fluidity helps you better understand your glazes and materials and solve all sorts of problems.
|Tests||Frit Softening Point|
|Tests||Glaze Melt Flow - Runway Test|
|Tests||Glaze Melt Fluidity - Ball Test|