Oxides, Carbonates and Ceramic Stains
Nearly every color in ceramics comes from a handful of metals. Here is what each one does, how much of it you need, why the same metal gives a different color in a different kiln, and how to choose between the raw oxide, the carbonate and the pre-made stain.
Oxides and carbonates are raw metals that react with the glaze and the kiln. Stains are pre-made and land where you aim them. Raw metals are cheaper, livelier and less predictable. Stains cost more and do what the label says.
Jump to a section
Raw metal reacts. A stain does not.
An oxide or a carbonate is a raw metal compound. Put it in a glaze and it takes part in the melt, reacting with everything else in there and with whatever the kiln atmosphere is doing. The color you get is a conversation between the metal, the glaze and the firing.
A ceramic stain has already had that conversation. A manufacturer combined metals with stabilizers, fired the mix, ground it back to powder and tested it. It arrives as a finished color that mostly stays the color it says on the jar.
So the choice is not really oxide versus stain. It is how much you want the kiln to have an opinion.
Which one to reach for
- You want a specific color and you want it again next year. Stain.
- You want depth, movement and surprise. Raw oxide or carbonate.
- You want a color the raw metals cannot make, like a clean pink, a bright yellow or a true purple. Stain, and it is the only answer.
- You are coloring a whole batch of clay or slip. Stain, because it disperses evenly and does not flux the body.
- You are making a wash to wipe into carving. Raw oxide, usually iron, because it is cheap and it moves.
- You are following a recipe. Use what the recipe says. Swapping an oxide for its carbonate changes the amount you need.
And the honest version: raw oxides cost a fraction of what stains cost, which matters more than any of the above once you are mixing in quantity.
Three things go wrong constantly
You used cobalt like it was iron
Iron shows up at 5 or 10 percent. Cobalt shows up at a quarter of one percent. Weigh cobalt on a scale that reads tenths of a gram, and treat any recipe calling for more than 2 percent with suspicion.
You swapped an oxide for a carbonate one for one
They are not interchangeable by weight. The carbonate carries less metal per gram, so you need roughly twice as much for the same color. Halve it going the other way.
You used a stain outside its range
Stains are formulated for specific temperatures, and plenty of them fade, shift or burn out past their limit. The bright pinks and reds are the fragile ones. Check the maker's chart before you commit a pot.
The three families
Everything in the color cupboard is one of these three, and knowing which is which answers most questions before you ask them.
- Metal oxides. A metal joined to oxygen, and nothing else. Red iron oxide, cobalt oxide, copper oxide, chrome oxide, manganese dioxide. The most concentrated form, the cheapest per unit of color, and usually the coarsest powder.
- Metal carbonates. The same metals joined to a carbonate group. Cobalt carbonate, copper carbonate, manganese carbonate. Lighter, finer and easier to disperse evenly, and they give off carbon dioxide as they fire, which matters if the glaze has already sealed over.
- Ceramic stains. Manufactured colors. Metals combined with stabilizers, fired, ground and tested until the result is repeatable. Often called Mason stains after the best known maker. Far more colors than the raw metals can reach, and far less drama.
- The ones that are not colorants at all. Tin oxide, zircon and titanium dioxide make a glaze white and opaque rather than colored. They matter because they change what every other colorant looks like.
- Rutile sits awkwardly between them. A mined mineral, mostly titanium with iron in it, so it both opacifies and colors, and it varies from bag to bag in a way refined chemicals do not.
- Price is a real factor. Raw oxides are cheap. Stains cost several times more per pound, and the cobalt-heavy and cadmium-based colors are the expensive end of an already expensive shelf.
The sentence that saves the most confusion
Raw metals color a glaze by reacting with it. Stains color a glaze by being suspended in it. That is why a stain looks similar in almost any base and an oxide can look like a different material entirely depending on what it lands in.
The seven that do most of the work
Percentages are of the dry glaze batch, and they are starting points for testing rather than promises.
Iron. 1 to 10 percent
The most used colorant in the history of pottery and the most forgiving. Pale amber at 1 or 2 percent, warm brown through the middle, saturated near black at 10 and above. Red iron oxide is the usual form. Yellow iron is the same metal, finer and softer. Iron is also the metal behind celadon and tenmoku.
Cobalt. 0.25 to 2 percent
By far the strongest, and the one that ruins tests. A quarter of one percent already reads blue. One to two percent is a full saturated blue. Past three or four it goes almost black and can start causing surface faults. Pure cobalt blue can read harsh, so many recipes soften it with a little iron, manganese or rutile.
Copper. 1 to 5 percent
Green to turquoise in oxidation, deepening as you add more. In reduction it makes copper red. It is volatile at high temperatures and can travel across the kiln onto other people's pots, which is a real consideration in a shared studio.
Manganese. 2 to 8 percent
Browns, purples and blacks, and largely indifferent to the kiln atmosphere. It gasses as it decomposes, which can pinhole a glaze that has already sealed over it. It is also the one with the most serious dust warnings on the jar.
Rutile. 3 to 10 percent
Not a strong color on its own. What it does is break the surface up into streaks and variegation, especially over an iron-bearing base. Around 3 to 5 percent gives subtle movement; 8 to 10 gives a heavily broken surface. Mineral, so it varies between bags.
Chrome. 0.5 to 2 percent
Green, and a troublemaker. With zinc it turns brown. With tin it makes pink, which is either the effect you wanted or a ruined white glaze on the next shelf, because chrome fumes at the top of our range and travels.
Tin. 5 to 10 percent
Not a color. An opacifier, and the one that gives the cleanest, brightest white of any of them. It is the basis of majolica and delftware. Expensive, and it is half of the chrome and tin pink problem, so keep tin glazes away from chrome ones.
Stains. 5 to 15 percent
Needed in much larger amounts than the raw metals, because most of the jar is stabilizer rather than metal. That is the trade for predictability. Always check the maker's temperature range first.
Why the ranges are so wide
The same percentage of the same colorant looks different in a glossy glaze and a matte one, and different again with zinc or magnesium in the base. A number from a book is where you start a line of tests, not where you finish. Fire five tiles at five percentages and you will learn more in one load than in a year of reading.
Oxide or carbonate, and how to swap
Cobalt sits on the shelf twice. So does copper and manganese. They are not the same thing by weight, and swapping one for the other without adjusting is a common way to ruin a batch.
- The carbonate is weaker per gram. Part of its weight is the carbonate group, which burns away. As a working rule, use about twice as much carbonate as oxide for the same color, and about half as much oxide as carbonate going the other way.
- The carbonate disperses better. It is a finer, lighter powder, which means fewer specks and a more even color. For most studio work that alone decides it.
- The oxide is more concentrated and cheaper to ship color in. If you are mixing in quantity and can grind and sieve properly, it goes further.
- Carbonates gas as they fire. They give off carbon dioxide on the way up. If a glaze has already melted and sealed over the top, that gas has to escape through it, which shows as pinholes or blisters.
- Carbonates are more soluble in water. In a bucket sitting for months that can slowly change how the glaze behaves, which is a reason to label and date everything.
- Not every metal comes both ways. Chrome and tin are used as oxides. Rutile is a mineral with no carbonate equivalent. Iron carbonate exists and is rarely used, because it decomposes unpredictably.
If a recipe just says cobalt
Assume it means the carbonate, because that is what most studio recipes are written with, and test a small batch before you trust it. The difference between 1 percent cobalt oxide and 1 percent cobalt carbonate is the difference between a deep navy and a mid blue.
When a stain is the right answer
Stains are not a beginner's shortcut. They are a different tool, and for several jobs they are the only tool.
- Colors the raw metals cannot make. Clean pinks, bright yellows, true purples, crimson. There is no oxide that gives you those at our temperature. If that is the color you want, a stain is the whole answer.
- Repeatability. The same jar, the same base, the same firing gives the same result. For anything you sell as a set, or anything you want to make again in a year, that is worth the money.
- Coloring clay and slip. Stains disperse evenly through a body without fluxing it the way a strong raw metal does, which is why colored porcelain and nerikomi work is nearly all stain-based.
- They need more of them. Five to fifteen percent, against a fraction of a percent for cobalt. Most of the jar is stabilizer, and you are paying for the stability.
- They have temperature limits, and they are real. Many stains are formulated for a range and fade, shift or burn out past it. The pinks, reds and oranges are the fragile ones. Check the chart, not the color of the powder.
- Some still interact. A few stain families dislike zinc, and some pinks need calcium in the base to develop. The jar is predictable; it is not immune to chemistry.
Oxidation and reduction
Oxidation means plenty of oxygen in the kiln. Reduction means starving it, so the fire pulls oxygen out of the materials themselves. Every kiln at The Clay Hole is electric and fires in oxidation, so the first column is what you actually get here.
Iron, the biggest difference
Oxidation: ambers, warm browns, near black at high percentages. Reduction: the same iron gives the cool greens and grays of celadon. Two completely different traditions from one metal.
Copper, the famous one
Oxidation: green and turquoise. Reduction: copper red, ox blood, the color Chinese potters chased for centuries. It needs the right base glaze and sustained reduction, and it is still difficult even then.
Cobalt, the reliable one
Blue in both. In a high-magnesium base it can shift toward purple in oxidation and pink in reduction, but for ordinary purposes cobalt does not care what the kiln is doing. That reliability is why it is in everything.
Manganese and stains
Both largely indifferent. Manganese gives browns, purples and blacks either way. Stains are built to be stable, so atmosphere is rarely the thing that breaks them. Temperature is.
Rutile
Oxidation: tan and amber with movement. Reduction: bluer and more broken, the effect people call rutile blue, driven by the iron riding along with the titanium.
What this means for a recipe you found
Most published glaze recipes come out of cone 10 reduction, because that is where a great deal of studio pottery writing comes from. Read what atmosphere and cone a recipe assumes before you decide it failed.
How much to use
Colorant percentages are worked out on the dry weight of the glaze batch, added on top of the recipe rather than counted inside it.
- Work out the dry batch weight first. Add up the recipe as written, ignoring water. That total is what your percentage refers to.
- Multiply. A 2000 gram dry batch at 2 percent cobalt carbonate is 40 grams of cobalt carbonate, added on top of the 2000.
- Use a scale that reads tenths of a gram for anything under 2 percent. On a small test batch, cobalt at half a percent can be under a gram, and a kitchen scale rounding to the nearest gram will double or erase it.
- Test a line, not a point. Mix one base and split it into five, at five percentages. One firing tells you where the color actually turns.
- Write it down while you weigh it, not afterward. A result you cannot reproduce is not a result.
The rough ladder, for a quick sense of scale
Cobalt from a quarter of a percent. Chrome and tin from half a percent. Copper from one. Iron from one to ten. Manganese from two to eight. Rutile from three to ten. Stains from five to fifteen. If a recipe puts a number wildly outside its band, check it before you weigh it.
Handling them safely
These are fine metal powders. Fired into a glaze and properly melted they are stable and ordinary. Loose in the air they are not, and the risk is entirely in the dry handling.
- A respirator, not a dust mask. Particulate filters, fitted properly, on before the lid comes off and kept on until the powder is wet and the bench is clean.
- Weigh over a tray, not over the floor. Spilled dry colorant becomes airborne every time somebody walks past it.
- Wet cleanup, always. Sponge and rinse. Never sweep or dry-wipe a bench that has had dry colorant on it, and never use compressed air.
- Manganese deserves extra care. It carries the strongest warnings of the common colorants. Treat it as the one you are strictest about.
- Gloves for mixing, and wash your hands before you touch your face, your phone or your lunch.
- Label everything. An unlabeled jar of gray powder is a problem for whoever finds it next, and in a shared studio that is somebody else.
What goes wrong, and what caused it
Everything came out almost black
Too much colorant, and it is nearly always cobalt or manganese. Go back to a quarter of what you used and test a line rather than a single tile.
Dark specks through an even color
A coarse oxide that never dispersed. Sieve the batch, or switch to the carbonate, which is finer and made for exactly this.
Pinholes and blisters
A carbonate gassing out after the glaze had already sealed. Less of it, a slower climb, or a hold near the top of the firing. Manganese does this most.
The white glaze next to yours went pink
Chrome from somebody's glaze met tin in somebody else's. Chrome fumes and travels in a shared kiln. It is the one interaction worth mentioning to whoever is loading.
The stain faded or vanished
Fired past its range. Stains are formulated for a temperature window and the brightest colors have the narrowest ones. Check the maker's chart against the cone you actually fire to.
The same percentage looks wrong in a different base
That is the raw metals behaving normally. Colorants react with what is around them, so zinc, magnesium, tin and the glossiness of the base all change the result. Retest in the base you are actually using.
The ones that come up at the tables
Can I just swap cobalt oxide for cobalt carbonate?
Not gram for gram. The carbonate carries less metal, so you need roughly twice as much of it for the same blue, and about half as much oxide if you are going the other way. Test a small batch either way, because the two also disperse differently.
Why is cobalt so expensive?
Supply and demand well outside pottery; the same metal goes into batteries. The consolation is that you need so little of it. A quarter of one percent already reads blue, so a small jar lasts a long time in a way a jar of iron does not.
Are these safe once they are fired?
In a properly melted, stable glaze, yes, and that is how the entire industry works. The risk is in the dry powder before it goes in. The other half of the answer is that a badly fitted or under-fired glaze can leach, which is why food surfaces want a tested glaze rather than an experimental one.
What is a Mason stain?
Mason is a manufacturer, and the name gets used generically the way people say Kleenex. A ceramic stain from any maker is the same idea: metals combined with stabilizers, fired, ground and tested so the color is repeatable.
Can I use these to color clay rather than glaze?
Yes, and stains are usually the better choice for it. They disperse evenly through a body and do not flux it the way a strong raw metal can. Expect to use more than you would in a glaze, and expect the fired color to be softer than the wet color.
Why does my copper glaze look different every firing?
Copper is one of the least stable colorants there is. It reacts to the base glaze, to temperature, to how full the kiln is and to what is firing next to it. That instability is exactly why people love it and exactly why it is frustrating.
What is an opacifier?
Something that makes a transparent glaze white and opaque without coloring it. Tin, zircon and titanium are the usual three. They matter for color work because they change what every colorant in the same glaze looks like, usually making it paler and softer.
Do I need all of these?
No. Iron, cobalt carbonate and rutile will take you a very long way, and iron alone covers more of the historical range of pottery than the rest of the shelf combined. Add the others when a specific result asks for them.
Where to go next
Making and using pottery washes
The simplest way to put these metals on a pot, with a calculator that weighs out any batch you want.
Read it →Oxide and carbonate washes
More recipes and more on how each metal behaves once it is on a surface rather than in a glaze.
Read it →Making and logging glaze test tiles
The line test is the method behind every percentage on this page. This is how to run one properly.
Read it →Mixing your own glazes
Where these colorants end up, and how a percentage of a dry batch actually gets weighed.
Read it →Underglaze techniques, rules and firing
Stains in their most convenient form: already blended, already suspended, ready to brush.
Read it →Clay Hole University
Every lesson and guide in one place, from first day at the wheel through firing and glaze chemistry.
Start there →Seven jars and a scale
Almost every glaze you have ever admired came out of this short list of metals. Not a secret ingredient, not a proprietary blend. Iron, cobalt, copper, manganese, rutile, chrome, tin, in different amounts, in different bases, in different kilns.
Which means the distance between reading about them and using them is one scale, a few small jars and the patience to fire a line of test tiles instead of a single hopeful pot. Start with iron. It is cheap, it is forgiving, and it has more range than anything else on the shelf.
Raw metals react with the glaze and the kiln. Stains are pre-made and stay put. Cobalt from a quarter of a percent, iron from one to ten, stains from five to fifteen. Carbonate needs about double the oxide.
If you want the full breakdown, every percentage and what changes between the two atmospheres, switch to the Comprehensive Version at the top of the page.
Which version would you like to print?
Videos come out as a labeled box with the title and channel to search for, since paper does not play them.