Science Page 8


Optical Microscope Mn Glaze 700x Matt Fiske
Image: Manganese crystals, 700x magnification, Matt Fiske

We may use manganese in our glazes as the carbonate, dioxide or manganese (III) oxide. The carbonate breaks down to the dioxide at around 200-400C, the dioxide to the (III) oxide at 530-650C, and the trioxide to Mn3O4 at 700-900C. This end result is actually a combination of (II) and (III) oxide, so can be written as manganese (II, III) oxide or Mn2O3.

Many potters claim that the kiln gives off manganese fumes, and this isn't helped by a highly misleading article on Digital Fire that is just supposition and has no measurements or other facts behind it, other than that something affected the author's health when she was making ceramics.. In all the analyses of kiln fumes I've seen, manganese levels are minimal or zero, way below any toxicity concerns. Mn3O4 has a melting point of 1567C, and a boiling point of 2847C, so unless global heating turns out to be much worse than predicted we can forget about manganese fumes. Inhaling the dust is a different matter.

Anyway, on to the glaze. There hasn't been that much research on manganese in silica glasses/glazes, and what there is doesn't focus on one aspect, so this is more of a collection of notes than a coherent study.

Manganese can have a valency of anywhere from +2 to +7. 2 and 3 are most common, but in glazes it can also occur as +4, acting directly as a glass former. The higher valencies form at higher temperatures, and are unstable. However at temperatures above 1000C the manganese forms as Mn(II)2 Mn(III) O4 - on cooling this breaks up and the Mn atoms take up their roles in the glaze.

Mn2+, with a coordination number of 6, acts as a network modifier in silica glasses. It gives a brown colour.
Mn3+ forms as a tetrahedron or octahedron in silica glasses, and so as a glass former and/or as a modifier, but only as an octahedron in boron glasses. The violet-amethyst colour of Mn3+ shifts towards blue whenever we use ROs and, to a lesser extent, R2Os of higher atomic weight, or towards red for a lower weight.
Mn4+ forms a tetrahedron, and acts as a glass former, but is only present in small amounts if at all

The ratio of (II):(III) varies with the glaze chemistry. For example, in Al-free glasses, only about 3% of the Mn is present as Mn(III), the rest having been reduced to Mn(II). But as the Al increases, the proportion of Mn(III) increases somewhat.

Saturated Mn Glazes Matt Fiske
Image: Saturated manganese glazes by Matt Fiske

If we have other multi-valent elements in the glaze, such as Cr, Ce, V, Cu, Sb, Fe or Sn, then there is generally a change of oxidation between any 2 elements. With Mn, all the above elements except for Cr will cause Mn to reduce to a valency of 2, whilst they are oxidised. This is used by glassmakers to remove the green tint from iron in glass - by adding the right amount of Mn it is reduced to Mn2+ and the iron oxidised to Fe3+, and then the colour spectra of the two cancel out to form a clear glass.

A pink stain can be produced with alumina and manganese, sometimes aided by boron or phosphorous. It re-oxidizes to brown on cooling unless covered by a glaze.

If you want to try something different, a glaze recipe of 15-30% MnO2, 60-75% potash feldspar, and small amounts of silica, kaolin, whiting, RIO and cobalt, will grow silvery lustrous crystals when fired in reduction. Both John Conrad and Matt Fiske have done a lot of work in this area.

In high lead glazes (e.g. 70 PbO, 30SiO2, 10 MnO2 by weight) we get various Mn-Pb-Si crystals forming; adding whiting or dolomite introduces Ca and/or Mg to the crystals, sharing the position of Mn. Results are complex, highly dependent on the firing temperature as well as the recipe.

In a boro-silicate glass with low levels of R2O (typically <5% molar) we get phase separation into Mn-B and Si phases. The Mn helps the B form tetrahedra, creating clusters of Mn-B.

So, like all multi-valent materials we use, manganese has a complex and interesting chemistry, even if not as fully explored as others like copper and iron, and without as wide a range of colours.


Zr Glaze Structure
Image: Zr in a boro-silicate glaze. Zr, blue octahedra; Si, yellow tetrahedra; B, green triangles and tetrahedral; Al, red tetrahedral; O, grey balls; Na, purple crosses; Ca, green crosses.

First, let's clarify the naming - potters can be quite sloppy over this. Zirconium (Zr) is the element, named after the mineral zircon (ZrSiO4) which is it's primary source - the name comes from the Persian word zargun, meaning gold coloured, which became Zircon in English via German. This is zirconium (IV) silicate. It may be sold under brand names such as Zirc-5 or Ultrox. There's also zirconium dioxide (ZrO2), or zirconia, but this is significantly more expensive than zircon, so is seldom used as a raw material, but forms during the firing of the glaze.

Zircon's high melting point means it won't melt directly into the glaze, but it forms a eutectic with Na and other fluxes that enable it to melt. However it's possible that if a lot of zircon is added to the glaze it won't all melt, but just be an inert inclusion in the glaze, acting as an opacifier.

In the glaze matrix, Zr cannot act as a glass former on its own, as it is too large, about double the diameter of Si. Instead, it forms ZrO6 octahedra with a charge of -2, with a balancing charge from a couple of alkali metals. Four O's link to adjacent tetrahedra in the glaze matrix, leaving two Os on their own; with alkali earths it forms ZrO7.

In a borosilicate glaze, Zr is most commonly attached to 4 Si and 2 B tetrahedra. The attractiveness of Zr to charge compensators falls between Al and B, so Zr will use whatever is left over from Al, and less will be available to B. This means that B will often be less able to form tetrahedra, increasing phase separation between B and Si. As always, the alkali metals are preferred for charge compensation over the alkali earths.

In a Bristol glaze, varying the ratio of ZrSiO4:ZnO from 3.8 to 67. The base UMF was KNa2O 0.14, CaO 0.6, MgO 0.16, ZnO 0.1, Al2O3 0.2, Si 2.24, ZrO2 0.165 and only ZnO and ZrO2 were changed. Whilst maximum whiteness was found at a ration of about 13, and maximum hardness at about 30, maximum strength was at 3.8.

Image: zircon structure. Blue Zr, orange Si, red O

As the availability of charge balancers decreases, e.g. by increasing the amount of Zr in the glaze, it moves from ZrO6 to ZrO7 and then ZrO8. Also, the Zr atoms move together, with the beginnings of phase separation, and it bonds to the Si and Al tetrahedra along the edges instead of the usual bonding on corners. These are the precursors of it being over saturated, and crystalising out of the glaze. Interestingly, tests indicate that the presence of boron increases the amount of Zr that can be dissolved into the glaze - I haven't found an explanation for this yet.

Zr greatly helps increase the durability of glazes - adding a small amount, so it doesn't act as an opacifier,is worth doing to just about all functional glazes, as it makes it harder for the attached Si tetrahedra to be leached out. The exception to this is on pieces that may stay damp for a long time, such as pieces going outdoors. In this environment, all glazes form a porous surface layer where material has been leached. In general, some of the leached Si moves from this layer back into the glaze, but Zr prevents this from happening, so over the longer term the glaze is more prone to corrosion.

Replacing Si with Zr increases the melting point of the glaze, makes it more viscous, and reduces the coefficient of thermal expansion. This, combined with an increase in elasticity of the glaze, makes the glaze less prone to crazing. It also has a high refractive index, so can make glazes appear brighter (though not as bright as a lead glaze).

Zircon Crystals In Glaze
Image: zirconia crystals in a glaze

So far we've talked about Zircon as part of the glaze matrix. But a lot of the time we want it to act as an opacifier, so we add more Zircon than can be held in the glaze. It then acts as a nucleating agent, and crystals are formed. These act as an opacifier not because they are white, but because they are a different refractive index to the glaze body. Each time light travels between the glaze and the crystal, it is refracted to a different direction, and so ultimately a minimal amount of light passes through the glaze.

The level at which zircon separates out from the glaze depends on the glaze chemistry. One set of tests in glasses found that this occurred from 6% molar Zr, with zirconia (ZrO2) and wollastonite (CaSiO3) forming; another found the formation of anorthite (CaAl2Si2O8) and diopside (MgCaSi2O6) in glazes. A comparison of Na, Li, Ca, Mg and Zn showed that Na was less likely to trigger crystallisation than the others. Crystallisation is more likely when there is a shortage of alkali metals or alkali earths to act as charge balancers for the Zircon, as then it cannot form tetrahedra in the glaze. The zirconia crystals form at around 900 - 950C.

Zircon crystals are extremely hard, much harder than metal cutlery, and when there is extensive crystal growth the crystals can protrude above the glaze surface. This is then the ideal condition for cutlery marking. Formation of zirconia and a silicate created a harder, tougher surface than just the zirconia.

Both zircon and zirconia are very hard. In moderate amounts they significantly toughen the glaze, but with too high an amount they may lie on the glaze surface, causing cutlery marking.

Zircon is also used extensively in making ceramic pigments and stains, with small amounts of Fe, V, Pr, Cd, S, or Se replacing Si or Zr in the crystals, producing the pigments. But that's another story.

R2O3 - Alumina and other intermediates

Al2O3 Vs R2O
Al2O3:R2O from Glazy recipes where there are no other R2O3s.

We're told that glazes need aluminium oxide (alumina) to toughen them up, but nobody really explains how that works, so I thought I'd look at that today.

On its own, silica forms a very strong and durable glass, made of tetrahedra of oxygen atoms with an Si atom in the middle of each, all joined at the corners with strong covalent bonds. But the melting point is too high, so we add fluxes to bring it down.

But the Na, K and Li in the feldspar have a charge of +1, and readily bond to the oxygen in the silica tetrahedra. This means that that corner of the tetrahedron can't bond to anything else. This breaks up and weakens the glass structure, so is not good news.

Now let's look at R2O3s like alumina. These cannot form a tetrahedron, as the metal just has a charge of +3, so is limited to a flat triangle. In this form they cannot participate in the silica network, as they are the wrong shape - though some, like Boron, can form their own glass structure out of just these triangles, but that's going off at a tangent. Anyway, the R2O3s will be more stable if they could get locked in to the silica network, so they get a bit cunning and buddy up with an Na or K or Li to give them the extra +ve charge needed to form a tetrahedron, and so can participate in the silica network.

This is great news all round, as the Al gets locked in to the silica network (otherwise it would act as a modifier or crystalise out), and it also hoovers up a lot of the alkali metals that would otherwise break up the silica matrix. So, with a few simplifications, for maximum glaze stability we should say that the number of alkali metal atoms should equal the number of aluminium or other R2O3 atoms. Too much alkali metal and it will break up the glaze matrix; too much R2O3 and it will separate out into a separate glass, or break up the glaze matrix, or crystallise out.

Other factors come into play, such as the melting point of the glaze, or achieving a desired appearance, and the alkali earths, but this is a good starting point.

Alkaline Earths

Al Si Flakelet
Image shows a 2-d view of a flakelet surrounded by ROs.

After last week's recap on the intermediates and alkali metals, I thought I'd do a recap on the ROs, such as the alkaline earths.

The first way in which they may be used is in glazes where there is a deficit of R2Os - although the intermediates much prefer to pair up with alkali metals, once used up they'll go for the ROs, even though they aren't as good a fit. With ROs having a charge of +2, it may be thought that one R could cover two Al, but because of the poor fit that isn't the case. The proportion is variable, but one paper found a ratio of 1.2Al:1R.

The fragments of glaze matrix cannot extend for ever, but form flakelets, typically either of Si or Si+Al+alkaline metal. On the flakelet surfaces the oxygen atoms will have one bond to the Si or Al, and the other unattached. These are where the ROs prefer to go, and as they have 2 bonds they join together two tetrahedra in the glaze, thereby strengthening the glaze (though not as much as adding on more Si or Al, as the RO uses weak ionic bonds, whereas Si and Al use the stronger covalent bonds).


SnO 1080x1350

On our glaze materials shelf we have SnO2, stannic oxide, where the tin has a charge of +4. But in glazes it may turn into SnO, stannous oxide. This happens if we fire in reduction, or at high temperatures it decomposes to SnO, though at kiln temperatures this doesn't happen completely.

Glaze composition also affects the ratio of SnO:SnO2; increasing the alkali metals shifts the balance towards more SnO2, whereas more alumina increases the proportion of SnO.

Whilst SnO is readily soluble in silica, boron and phosphor glazes, SnO2 has very low solubility (less than a few % molar, though this increases as the alkali metals increase), and is slow to dissolve. This low solubility makes it an effective opacifier, with a glossier result than zircon, though it turns pink if chromium is present.

But there's more to tin than being an opacifier. It is used with gold, copper and selenium in forming red glazes. In copper red glazes, the initial reduction firing forms SnO, but this later reoxidises to SnO2, taking the oxygen from any nearby copper oxide atoms, reducing them to copper metal. It's role with gold and selenium isn't yet fully understood. SnO is also a lot more volatile than SnO2, and a significant amount will vapourise off in the kiln.

SnO forms a tetrahedron, with Sn at the apex and four O at the base; these are then stacked, as shown in the lower image. Or you can visualise it as a stretched cube of 8 oxygens enclosing 4 Sn, with the Sn nearer to one end rather than in the middle. The tin atom is polarised, with the top having a negative charge; from one side it looks like it is Sn, and the other looks like the Sn4+ ion. The amount of SnO that can be incorporated into the glass/glaze is very high, e.g. glasses can be formed with 70% Sn and 30% Si, for similar reasons as for lead, that occurs just below Sn is the periodic table.

At concentrations of SnO of up to 20% molar (well above most glazes), it normally acts as a network modifier, but above this it acts as an intermediate. Also, in boron glazes, it links up with the boron to let the boron form a tetrahedron, and so enter the silica glaze matrix.

A couple of things can happen next. First, the flakelets are separated by rivers of left-over bits and pieces, and they may end up here. But if there is a significant excess the ROs are also likely to crystalise out, giving us everything from matt glazes to large crystals.

Because we don't know the surface areas of the flakelets, we can't tell how many ROs are needed to bond to all those loose oxygen bonds on the surface of each flakelet (I suspect that flakelet sizes, and so the total surface area, is at least as dependent on the cooling schedule as the glaze chemistry). And (if there's a deficit of alkali metals) we only know that the ratio of Al:R for Al tetrahedra without an alkali metal must be between 1:1 and 2:1. So, unlike with the alkali metals, we cannot specify a minimum proportion of RO to ensure a stable glaze. The best we can probably do is look for the first signs of crystalisation to tell us that the minimum required level has been exceeded, and then (if we want to) reduce it by a small amount to give a non-crystaline gloss glaze again.

SnO2 is more complex than SnO. Where the ratio of R2O:R2O3 is less than 1, the solubility of Sn in the glaze is low, and the Sn is linked to 6 oxygens, as shown in the image. But at higher levels of R2O there is spare R2O, and this combines with Sn, Si and other glaze components to form complex structures, increasing the solubility of Sn, with it linking to 4 or 6 O. There is also some evidence that these can form clusters within the glaze. One paper found that adding up to 5% SnO2 to a glass increased network connectivity, making the glass denser, more stable and with a higher melting point.

I've mentioned before how one glaze material can reduce or oxidise another. This happens because they pass or share electrons between each other, changing the number of oxygen atoms that they can bond with. And they will do this to increase the overall stability of the glaze. But how do we know what may happen?

Ellingham 2

One way of determining this is the Ellingham diagram, shown in the image. Simplifying a bit, each line on the graph shows an oxide. The temperature is shown horizontally, and the Gibbs free energy is on the vertical axis. Without going into the Gibbs thing, we can use this diagram in a couple of ways. First, for a given element, the most stable oxide is at the bottom, and the greater the separation, the more stable that oxide is - e.g. Al2O3 is much more stable than Al2O. Secondly, for any line the oxide on the line will be formed by oxidising reduced forms, and oxides above this line being reduced. So, for example, if we have copper and tin in the glaze, the following may occur:
SnO at temperatures up to about 900C lies above CuO and Cu2O, so no reaction is possible. Above 900, SnO is below CuO, so Sn will go to SnO and CuO will go to Cu (or Sn + CuO -> SnO + Cu).
SnO2 is below Cu2O and CuO, so SnO + Cu2O -> SnO2 + CuO and SnO + CuO -> SnO2 + Cu

There are some simplifications in applying Ellingam diagrams to glazes, so the outcomes should be seen as likely to happen rather than definitely. Also, I've not covered the difference of the kiln being in oxidation or in reduction here. But you can see how one oxide can reduce or oxidise another.


Wine Bottle From Korea, 17th Century, Porcelain , Dayton Art Institute
Cobalt has been used as a pigment in glass and ceramics for about 4500 years, but here we're concentrating more on its structural behaviour than in its ability to create pretty blue, purple or pink pots, so we won't be looking at the structure or colour of the many cobalt-containing stains..

The name, if you're interested, comes from the German "kobold", a type of goblin from pre-Christian times, one group of which lived underground in mines. When first identified as an element by George Brandt it was called Kobolt after the poor quality ores that gave off arsenic fumes when smelted, associated with the kobolds.

We introduce cobalt as the oxide (normally Co3O4) or the carbonate (CoCO3) - both of which break down to CoO by somewhere above 800C. In this form the cobalt has a valency of 2.

As CoO, it acts as a network modifier in the glaze, though if too much is added it may form a cobalt alumina spinel. When there are 4 oxygen atoms around each Co2+ ion it forms a tetrahedron, acting as a glass former, and one gets a blue. With 6 O neighbours it forms an octahedron, acting as a network modifier, and we get pink or red colours. Formation of the 6 is encouraged in the presence of boron or phosphorous, but it is also affected by firing schedules. We also get blue Co3+ ions, which also form tetrahedra and act as glass formers.


F Phase Separated
Fluorine phase separation

Collapsed Clinker 1080
A vase with a glaze of 50:50 cryolite and nepheline syenite that split in half in the kiln

Fluorine is a halogen, lying to the right of oxygen in the periodic table, with a charge of -1. It is highly reactive, and we see this in it's tendency to replace oxygen in the glaze. With its charge of -1, it can only connect to one silica atom, and so that reduces the connectivity of the glaze, and reduces the melting point and viscosity.

Historically, small amounts were added to enamels to get them to flow better and remove bubbles when fired. But that's not all that happens. It can also act as a seed for crystal growth, and in glassmaking, it is added to make opalescent glass. Also, in my experience, it can severely attack some clay bodies if used extensively - the 2nd image is a thin walled vase that split in two during the firing, with a glaze of 50:50 nepheline syenite and cryolite. The fired clay had no more strength in it than a biscuit.

So what's the chemistry?

In the glaze, F prefers to bond to Al, forming structures with 4, 5 or 6 F to one Al. Its next preference is to bond to alkali earths as RF2, and then to alkali metals as RF, and finally to Si to form SiF4. Both SiF4 and RF compounds have a low enough boiling point to be given off as fumes.

The solubility of F depends greatly on the glaze chemistry. In the absence of Al and the alkaline earths it is very low, less than 2% by weight. With these present, it can be 16% or more, especially if the glaze is high in Al and alkaline earths, and they are of heavier molecular weight.

With low amounts of F, it just mixes in with the rest of the glaze - and some will bond with Si, giving off SiF4 fumes. But, as more is added, we get a phase separation occuring, as shown in the first image. The glaze separates into an SiO2-Al2O3 phase (with enough R2O to match the Al), and an AlF3-RF2 phase (where R is an alkali earth); the more F in the glaze, and the greater the proportion of Ba to Ca, the greater the phase separation. This means that the F and Si cannot intermingle, and so SiF4 and RF are not produced. Thus, to reduce the level of fluorine fumes, we need to increase the amount of fluorine in the glaze, to trigger the phase separation.

There are a number of fluorine glazes in the book Dry Glazes, mostly from John Chalke who used cryolite extensively. But be aware that high fluorine glazes will give off fumes that will significantly shorten the life of your elements, as well as etching the glass in your windows.

SiF4 is highly corrosive to the skin and eyes, so if using in any quantity ensure that you have good ventilation. Safe levels are 2.5mg/cu.m. averaged over a 10 hour day, or 5mg/cu.m. averaged over 15 minute exposure, so about 1/20th that for carbon monoxide for comparison.


Nickel Glass Colours
It is perhaps best known for giving greyish colours, but get the chemistry right and a much wider spectrum is available. Unfortunately nickel's colours have a reputation for being fickle - due to a lack of understanding of the chemistry. Because of it's colour variability, it wasn't often used as a primary colour source, but more often as a modifier for other colours, but that needn't be the case.

The colour and structure vary with the other glaze components. With K it forms pink to purple 4-fold tetrahedra, but with Li or P it forms yellow-brown to brown to red-pink 5-fold structures; with Na it is a brown mix of the two. Alkali earths also affect the colour, but to a lesser extent. Ba and Ca give brown; Mg green; whilst a mix of Ca, Mg and Zn gives grey. And glazes high in B or Mg or Zn give a blue to green 6-fold structure, with the Ni possibly going from Ni(II) to Ni(III). The grey comes from a mix of yellow and purple, but as the glaze gets thicker there's a shift towards purple, as red light isn't absorbed by either the yellow or the purple colours. The image shows some of these colours.

So why is this happening? As the field strength (given by ionic charge/(ionic radius squared)) of the other glaze components increases, the nickel shifts from 4- through 5- to 6-fold forms.

To put some numbers on this, here's data on an Al-free glass. In an K/Si glass, 4-fold dominated when K2O was 12-28% by weight; in an Na/Si glass, 5-fold dominated at Na2O 15-27%; and a mix of 4- and 5-fold when there was a mix of Na and K. In a K/Ca/Si glass with 10% K2O and 9-26% CaO, the 5-fold form increased as CaO increased; with CaO fixed at 10% and K2O varied, only the 4-fold form was found. In an Na/Ca/Si glass, only the 4-fold form was found.

If we now turn this into a glaze by replacing 15% of the silica by weight with alumina, there is a decrease in the 4-fold Ni. This shows that the Al is better at attracting charge compensating R2Os and ROs than Ni.

An example of the 6-fold form occurs in an Na/K/Ca/Mg/Si glaze with 7-20% MgO. Increasing MgO increased the 6-fold form at the expense of the 5-fold. Another is in high boron glasses, with <10% by weight of alkali metals.

At high temperatures much more of the nickel forms 4-fold tetrahedra, but with cooling the solubility of the Ni decreases and the excess Ni separates out into 5- and 6-fold forms - the slower the cooling, the more tetrahedra break down. The solubility of NiO in the solid glaze is low, around 1 - 5% by weight, much less than at high firing temperatures, so as the glaze cools the excess Ni often separates out into crystals at about 1100-1200C.

As you may have expected, the 4-fold form creates a more durable and stronger glaze than the 5- or 6- forms.

In reduction, some of the nickel oxide is reduced to metal, and the metal aggregates into ever larger clumps with time; the remaining nickel shifts towards the 4-fold form.

As an aside, the behaviour of Mg and Ni are very similar in the glaze. So replacing some Mg with Ni and looking at the colour will tell you what role the Mg is taking - if you recall, it can also act as a glass former.

Page 7 <