Glaze chemistry is learning what each oxide does in a fired glaze and the relative advantages and disadvantages of each material supplying it. The chemistry of a glaze is expressed in a manner similar to its recipe, except that the items are oxides and the amounts can be by weight (an analysis) or numbers of molecules (a formula).
Ceramic Oxide Periodic Table
All common traditional ceramic base glazes are made from only a dozen elements (plus oxygen). Materials decompose when glazes melt, sourcing these elements in oxide form. The kiln builds the glaze from these, it does not care what material sources what oxide (assuming, of course, that all materials do melt or dissolve completely into the melt to release those oxides). Each of these oxides contributes specific properties to the glass. So, you can look at a formula and make a good prediction of the properties of the fired glaze. And know what specific oxide to increase or decrease to move a property in a given direction (e.g. melting behavior, hardness, durability, thermal expansion, color, gloss, crystallization). And know about how they interact (affecting each other). This is powerful. And it is simpler than looking at glazes as recipes of hundreds of different materials (each sources multiple oxides so adjusting it affects multiple properties).
Glaze chemistry works best when changing material amounts, not material types
Two glazes, same chemistry, different materials. The glaze on the left is sourcing CaO from wollastonite, the one on the right from calcium carbonate. Thus the oxide chemistry of the two is the same but the recipe of materials sourcing that chemistry is different. The difference in this GLFL test for melt flow is an expression of how choosing different mineral sources to source an oxide can produce melting patterns that go outside what the chemistry suggests. The difference here is not extreme, but it can be much more. Glaze chemistry is relative, not absolute. It works best when you are changing material amounts, not material types. When you do introduce a very different mineral then you have a different system which has its own relative chemistry.
What can you do using glaze chemistry?
There is a direct relationship between the way ceramic glazes fire and their chemistry. Wrapping your mind around that and overcome your aversion to chemistry is a key to getting control of your glazes. You can fix problems like crazing, blistering, pinholing, settling, gelling, clouding, leaching, crawling, marking, scratching, powdering. Substitute frits or incorporate better, cheaper materials, replace no-longer-available ones (all while maintaining the same chemistry). Adjust melting temperature, gloss, surface character, color. Identify weaknesses in glazes to avoid problems. Create and optimize base glazes to work with difficult colors or stains and for special effects dependent on opacification, crystallization or variegation. Create glazes from scratch and use your own native materials in the highest possible percentage.
This cone 6 transparent looked good, but I still improved it alot
The green boxes show cone 6 Perkins Studio Clear (left) beside an adjustment to it that I am working on (right). I am logged in to my account at insight-live.com. In the recipe on the right, code-numbered G2926A, I am using the calculation tools it provides to substitute Frit 3134 for Gerstley Borate (while maintaining the oxide chemistry). A melt-flow GLFL test comparison of the two (bottom left) shows that the GB version has an amber coloration (from its iron) and that it flows a little more (it has already dripped off). The flow test on the upper left shows G2926A flowing beside PGF1 transparent (a tableware glaze used in industry). Its extra flow indicates that it is too fluid, it can accept some silica. This is very good news because the more silica any glaze can accept the harder, more stable and lower expansion it will be. You might be surprised how much it took, yet still melts to a crystal clear. See the article to find out.
Cutlery marking is directly related to the chemistry of the glaze
This is an example of cutlery marking in a cone 10 silky matte glaze lacking Al2O3, SiO2 and having too much MgO. Al2O3-deficient glazes often have high melt fluidity and run during firing, this freezes to a glass that lacks durability and hardness. But sufficient MgO levels can stabilize the melt and produce a glaze that appears stable but is not. Glazes need sufficient Al2O3 (and SiO2) to develop hardness and durability. Only after viewing the chemistry of this glaze did the cause for the marking become evident. This is an excellent demonstration of how imbalance in chemistry has real consequences. It is certainly possible to make a dolomite matte high temperature glaze that will not do this (g2571a is an example, it has lower MgO and higher Al2O3 and produces the same pleasant matte surface).
Exporting insight-live recipes to a CSV file
An Insight-live page displaying four cone 6 matte recipes. It has been exported to a CSV file which I have opened in my spreadsheet software. I then reorganized it to compare these 4 glazes and relate the chemistry to the melt flow tests.
Two stains. 4 colors. Will the guilty oxide please step forward.
We are looking at two pairs of samples, they demonstrate why knowing about glaze chemistry can be so important. Both pairs are the same glazes: G2934 cone 6 matte and G2916F cone 6 glossy. The left pair has 5% maroon stain added, the right pair 5% purple stain. The red and purple develop correctly in the glossy but not the matte. Why? The Mason Colorworks reference guide has the same precaution for both stains: the host glaze must be zincless and have 6.7-8.4% CaO (this is a little unclear, it is actually expressing a minimum, the more the CaO the better). The left-most samples of each pair here have 11% CaO, the right-most have 9%. So there is enough CaO. The problem is MgO (it is the mechanism of the matteness in the left two), it impedes the development of both colors. When you talk to tech support at any stain company they need to know the chemistry of your glaze to help.
2, 5, 10 and 15% calcined alumina added to Ravenscrag Slip
The Ravenscag:Alumina mix was applied to a buff stoneware fired at cone 10R (by Kat Valenzuela). Matting begins at only 5% producing a very dry surface by 15%. This "psuedo matte" surface is simply a product of the refractory nature of the alumina as a material, it does not disassociate in the melt to yield its Al2O3 as an oxide (as would a feldspar, frit or clay). The same test using alumina hydrate demonstrates that it disassociates somewhat better (although not completely).
L3617 Cornwall Stone substitute vs. real Cornwall Stone
These flow tests demonstrate how similar the substitute recipe (left) is to the real material (right). 20% Frit 3134 has been added to each to enable better melting at cone 5 (they do not flow even at cone 11 without the frit). Links below provide the recipe for the substitute and outline the method of how it was derived using Digitalfire Insight software. This substitute is chemically equivalent to what we feel is the best average for the chemistry of Cornwall Stone.
Employing a frit on unknown chemistry in your glaze recipe ..
Is like adding a dog of unknown breed to this team. How predictable is that going to be? It is like that with ceramic glazes. They fire the way they do because of their chemistry. Not knowing the chemical makeup of a key ingredient robs you of the single biggest tool to explain characteristics or issues or propose and adjustments, improvements or fixes.
The language of art will never adequately describe this issue or its solution
This is crazing. It is bad on functional ware. Shivering and leaching are also bad. And blistering. And other problems like devitrification and clouding compromise the visual appearance of ware. Science is needed to understand what these are, how serious they are and determine the cause and solution. And the language of science is needed to clearly describe all of this. And methodical testing to prove the solution. It is not rocket science, just simple ceramic science.
Example of a whole rock chemical analysis lab report
Powdered samples were sent to the lab. The numbers shown on this report are in percentage-by-weight. That means, for example, that 15.21% of the weight of the dry powder of Alberta Slip is Al2O3. Insight-live knows material chemistries in this way (whereas desktop Insight needs them as formulas). Some non-oxide elements are quantified as parts-per-million (these amounts are not normally high enough to take into account for traditional ceramic purposes). The LOI column shows how much mechanically and chemically bound water are gassed off during firing of the sample. The total is not exactly 100 because of inherent error in the method and compounds not included in the report.
Compare fired glaze melt fluidity balls with their chemistry and lights come on!
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).
Magic of Fire book
A book published by Tony Hansen. It explained why were need to think about materials (and the bodies and glazes made from them) as more than just powders. They have physical, chemical and mineralogical presences that do not take a lot of effort to understand. This was the first widely read book to show how, armed with this information and a knowledge of how to do glaze chemistry, readers could solve all sorts of problems. It showcased the real value of the oxide viewpoint in ceramics and explained how to use Digitalfire Insight and Foresight software in each scenario. From 2000-2014, the book was used as courseware in universities around the world. In 2015 the book was temporarily removed from distribution at digitalfire.com awaiting a new edition. Readers were reminded that all of the books content was available at the Digitalfire Reference Library.
Do you know the purpose of these common Ferro frits?
I used a binder to form 10 gram GBMF test balls and fired them at cone 08 (1700F). Frits melt really well, they do not gas and they have chemistries we cannot get from raw materials (similar ones to these are sold by other manufacturers). These contain boron (B2O3), it is magic, a low expansion super-melter. Frit 3124 (glossy) and 3195 (silky matte) are balanced-chemistry bases (just add 10-15% kaolin for a cone 04 glaze, or more silica+kaolin to go higher). Consider Frit 3110 a man-made low-Al2O3 super feldspar. Its high-sodium makes it high thermal expansion. It works in bodies and is great to incorporate into glazes that shiver. The high-MgO Frit 3249 (for the abrasives industry) has a very-low expansion, it is great for fixing crazing glazes. Frit 3134 is similar to 3124 but without Al2O3. Use it where the glaze does not need more Al2O3 (e.g. it already has enough clay). It is no accident that these are used by potters in North America, they complement each other well. The Gerstley Borate is a natural source of boron (with issues frits do not have).
Frits work much better in glaze chemistry
The same glaze with MgO sourced from a frit (left) and from talc (right). The glaze is 1215U. Notice how much more the fritted one melts, even though they have the same chemistry. Frits are predictable when using glaze chemistry, it is more absolute and less relative. Mineral sources of oxides impose their own melting patterns and when one is substituted for another to supply an oxide in a glaze a different system with its own relative chemistry is entered. But when changing form one frit to another to supply an oxide or set of oxides, the melting properties stay within the same system and are predictable.
A Limitation of the Seger Unity Formula
Mineral sources of oxides impose their own melting patterns and when one is substituted for another to supply an oxide a different system with its own relative chemistry is entered. An extreme example of this would be to source Al2O3 to a glaze using calcined alumina instead of kaolin. Although the formula may be exactly the same, the fired result would be completely different because very little of the alumina would dissolve into the glaze melt. At the opposite extreme, a different frit could be used to supply a set of oxides (while maintaining the overall chemistry of the glaze) and the fired result would be much more chemically predictable. Why? Because the readily and release their oxides the the melt.
2, 5, 10 and 15% alumina hydrate added to Ravenscrag Slip
Pure Ravenscrag Slip is glaze-like by itself (thus tolerating the alumina addition while still melting as a glaze). It was applied on a buff stoneware which was then fired at cone 10R (by Kat Valenzuela). This same test was done using equal additions of calcined alumina. The results suggest that the hydrated version is decomposing to yield some of its Al2O3, as an oxide, to the glaze melt. By 15% it is matting and producing a silky surface. However crazing also starts at 10%. The more Al2O3 added the lower the glaze expansion should be, so why is this happening? It appears that the disassociation is not complete, raw material remains to impose its high expansion.
Compare two glazes having different mechanisms for their matteness
These are two cone 6 matte glazes (shown side by side in an account at Insight-live). G1214Z is high calcium and a high silica:alumina ratio (you can find more about it by googling 1214Z). It crystallizes during cooling to make the matte effect and the degree of matteness is adjustable by trimming the silica content (but notice how much it runs). The G2928C has high MgO and it produces the classic silky matte by micro-wrinkling the surface, its matteness is adjustable by trimming the calcined kaolin. CaO is a standard oxide that is in almost all glazes, 0.4 is not high for it. But you would never normally see more than 0.3 of MgO in a cone 6 glaze (if you do it will likely be unstable). The G2928C also has 5% tin, if that was not there it would be darker than the other one because Ravenscrag Slip has a little iron. This was made by recalculating the Moore's Matte recipe to use as much Ravenscrag Slip as possible yet keep the overall chemistry the same. This glaze actually has texture like a dolomite matte at cone 10R, it is great. And it has wonderful application properties. And it does not craze, on Plainsman M370 (it even survived a 300F-to-ice water IWCT test). This looks like it could be a great liner glaze.
These two frits have one difference in the chemistry: Al2O3.
These two boron frits (Ferro 3124 left, 3134 right) have almost the same chemistry. But there is one difference: The one on the right has no Al2O3, the one on the left has 10%. Alumina plays an important role (as an oxide that builds the glass) in stiffening the melt, giving it body and lowering its thermal expansion, you can see that in the way these flow when melting at 1800F. The frit on the right is invaluable where the glaze needs clay to suspend it (because the clay can supply the Al2O3). The frit on the left is better when the glaze already has plenty of clay, so it supplies the Al2O3. Of course, you need to be able to do the chemistry to figure out how to substitute these for each other because it involves changing the silica and kaolin amounts in the recipe also.
The unexpected reason for this crazing can be seen in the chemistry
This liner glaze is 10% calcium carbonate added to Ravenscrag slip. Ravenscrag Slip does not craze when used by itself as a glaze at cone 10R on this body, so why would adding a relatively low expansion flux like CaO make it craze? It does not craze when adding 10% talc. This is an excellent example of the value to looking at the chemistry (the three are shown side-by-side in my account at Insight-live.com). The added CaO pushes the very-low-expansion Al2O3 and SiO2 down by 30% (in the unity formula), so the much higher expansion of all the others drives the expansion of the whole way up. And talc? It contains SiO2, so the SiO2 is not driven down nearly as much. In addition, MgO has a much lower expansion than CaO does.
What could make glazes grow these incredible crystals?
Closeup of a crystalline glaze by Fara Shimbo. Crystals of this type can grow very large (centimeters) in size. They grow because the chemistry of the glaze and the firing have been tuned to encourage them. This involves melts that are highly fluid (lots of fluxes) with added metal oxides and a catalyst. The fluxes are normally B2O3, K2O and Na2O (from frits), the catalyst is zinc oxide (alot of it). Because Al2O3 stiffens glaze melts preventing crystal growth, it is very low in these glazes (clays and feldspars supply Al2O3, so these glazes have almost none). The firing has a highly controlled cooling cycle involving rapid descents and holds (sometimes multiple cycles of these). Between the cycles there are sometimes slight rises. Each discontinuity in the cooling curve creates specific effects in the crystal growth. Thousands of potters worldwide have investigated the complexities of the chemistry, the firing and the infinite range of metal oxides additions.
Calcium carbonate and glaze bubbles
The two cone 04 glazes on the right have the same chemistry but the center one sources it's CaO from 12% calcium carbonate and ulexite (the other from Gerstley Borate). The glaze on the far left? It is almost bubble free yet it has 27% calcium carbonate. Why? It is fired to cone 6. At lower temperatures carbonates and hydrates (in body and glaze) are more likely to form gas bubbles because that is where they are decomposing (into the oxides that stay around and build the glass and the ones that are escaping as a gas). By cone 6 the bubbles have had lots of time to clear.
What happens when glazes lack Al2O3?
This happens. They are glossy, but lack thickness and body. They are also prone to boron blue clouding (micro crystallization that occurs because low alumina melts crystallize much more readily on cooling). Another problem is lack of resistance to wear and to leaching (sufficient Al2O3 in the chemistry is essential to producing a strong and durable glass). This is a good example of the need to see a glaze not just as a recipe but as a chemical formula of oxides. The latter view enables us to compare it with other common recipes and the very low Al2O3 is immediately evident. Another problem: Low clay content (this has only 7.5% kaolin) creates a slurry that is difficult to use and quickly settles hard in the bucket.
How to include stains in chemistry calculations in Insight
The simple answer is that you should not. The chemistry of stains is proprietary. Stain particles do not dissolve into the glaze melt like other materials, they suspend in the transparent glass to color it. That is why stains are color stable and dependable. In addition, their percentage in the recipe, not the formula, is the predictor of their effect on the fired glaze. Of course they do impose effects on the thermal expansion, melt fluidity, etc., but these must be rationalized by experience and testing. But you can still enter stains into Insight recipes. Consider adding the stains you use to your private materials database (for costing purposes for example).
Same high-iron glaze. One crystallizes and the other does not. Why?
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.
Glazes of the same chemistry: The fritted one melts better
It seems logical (and convenient) to just say that the kiln does not care what materials source the oxides in a glaze melt. Li2O, CaO, Al2O3, SiO2 are oxides (there are about ten common ones). The kiln just melts everything and constructs the glaze from the ones available. Right? Wrong! Things get more complicated when frits are introduced. Frits are man-made glasses, they melt much more readily than raw materials like feldspar. Raw materials are often crystalline. Crystals put up a fuss when asked to melt, often holding on as long as they can and then suddenly melting. Frits soften over a range and they start melting early. To illustrate: These two glazes have the same chemistry. But the one on the left sources sodium and alumina (Na2O3, Al2O3) from the 48% feldspar present. The other sources these from a frit (only 30% is needed for the same amount of Na2O3). The remainder of the recipe has been juggled to match the other oxides. The frit version is crystallizing on cooling (further testament to how fluid the melt is). What has happened here is great. Why? First, the chemistry has not changed (fewer firing differences). The frit has no Al2O3, it is being sourced from kaolin instead, now the slurry does not settle like a rock. Even better, silica can be added until the melt flow matches (might be up to 20%). That will drop the thermal expansion and reduce crazing. The added SiO2 will add resistance leaching and add durability. Frits are great! But you need to know how to incorporate them into a recipe using a little glaze chemistry.
A settling, running glaze recipe gets a makeover
The original cone 6 recipe, WCB, fires to a beautiful brilliant deep blue green (shown in column 2 of this Insight-live screen-shot). But it is crazing and settling badly in the bucket. The crazing is because of high KNaO (potassium and sodium from the high feldspar). The settling is because there is almost no clay. Adjustment 1 (column 3) eliminates the feldspar and sources Al2O3 from kaolin and KNaO from Frit 3110. The chemistry of the new chemistry is very close. To make that happen the amounts of other materials had to be juggled (you can click on any material to see what oxides it contributes). But the fired test reveals that this one, although very similar, is melting more (because the frit releases its oxide more readily than feldspar). Adjustment 2 (column 4) proposes a 10-part silica addition (to supply more SiO2). SiO2 is the glass former, the more a glaze will accept, the better. Silica is refractory so the glaze will run less. It will also fire more durable and be more resistant to leaching.
Why would a low fire transparent require four frits?
To get the needed chemistry to avoid boron blue clouding (calcium borate crystals). The one on the right clouds, the other does not. Why? Differences in the chemistry (as seen in my account at insight-live.com). G2931K, on the left, has greater Al2O3 (which impedes the growth of crystals), lower CaO (starves their growth) and more boron (for better melting). There is actually no practical way to adjust the recipe on the right (by supplying MgO with talc and fiddling with frit percentages) to achieve this. Frit 3124 lacks Na2O and B2O3. 3134 has excessive CaO and almost zero Al2O3. Talc does not melt well enough. But Frit 3249 supplies the needed MgO and has lots of B2O3 and low CaO. And Frit 3110 has low CaO and supplies the needed Na2O.
Out Bound Links
In Bound Links