Science Page 6
Alkali Metals
Let's look at sodium, potassium and lithium in a bit more detail.
First, melting. Sodium and potassium are normally added as feldspars, of formula R(AlSi3O8), where R is Na or K. In practice, all feldspars we use are a mix of sodium and potassium. They are the most common glaze fluxes. Other sources are nepheline syenite and Cornwall/Cornish stone (though the later is no longer mined, but a blend of other rocks). Or we can use frits, though these often also introduce other fluxes. For lithium, I use lithium carbonate, or there's spodumene and petalite.
All have low melting points - with a fair temperature difference between starting to melt and full melting. Lithium carbonate melts at 732C, and the lithium minerals at about 1000C. Then soda feldspar is melted by about 1050C, and potash by 1150C. The low melting points are important because many of our other materials don't melt at kiln temperatures, but instead dissolve into these early melters (others include lead, zinc, boron and bismuth).
Although often seen as interchangeable, another difference between the two feldspars is that once melted potassium is very stiff, whereas sodium creates a very runny glaze - reducing bubbles but also gluing the pot to the kiln shelf. Then on cooling both have very large coefficients of thermal expansion, so are very likely to cause crazing - volumetric coefficients are 16.9 for Na and 15.2 for K, with 5.0 for Li and most other materials around 4-5. Putting aside any aesthetic considerations, crazing weakens the pot, and also lets water permeate.
As you'll have gathered if you've been following these posts for a while, the alkali metals are also key to making the intermediates (Al, B, Ti etc) join in with the glass formers and make a stable glaze - they do this much better than the alkali earths. So even if we're firing to 1300C/C10 we need to have enough alkali metals to pair up with the intermediates.
Finally, the glossiness of the glaze depends on the Na:K ratio, with maximum gloss in one set of tests found at 1.4 - 1.6. As well as making the pot nice and shiny, it is also more resistant to staining and easier to clean.
Calcium
The white cliffs of Dover, made of chalk (aka Whiting) - often the first sight refugees get of Britain, when crossing the Channel from France
I've looked at the behaviour of the alkali earths in general, but now we can look at some of the differences between calcium, magnesium, barium and strontium, plus beryllium for those who like to be exotic. We'll give radium a miss!
First, a quick recap. All are fluxes, with a charge of +2. So, although their ionic bonds weaken the glaze matrix, they do link together two tetrahedra from the glaze matrix, so they don't weaken the glaze as much as the alkalis metals. Also, if you recall, the glaze is made of flakelets of glass surrounded by rivers of flux; the alkali earths tend to congregate on the outer boundary of the flakelets, with the alkali metals in the middle of the rivers, making the flakelets more stable. The other consideration is how they act with the intermediaries, which need an additional charge to enter the glaze matrix. Whilst the alkali earths can fill this role, they cannot easily attach to two intermediaries within the glaze, especially as two intermediaries don't fit well together, so it is a bit of a stretch. Thus this role is always first filled by the alkali metals, and the alkali earths only join in once they're all used up. Having double the charge, you may think it needs half the number of alkali earths, but in practice because they're a poor fit more than that are needed.
Calcium is primarily added to clays and glazes as whiting, CaCO3 (chalk or limestone for non-potters) or dolomite (mixed with magnesium), or wollastonite, CaSiO3. Wollastonite has the benefits of not giving off carbon dioxide (the carbonate decomposes to the oxide at 940C, and some glazes will have melted by then, resulting in bubbles in the glaze) and, being a silicate, doesn't need as much time or heat to join the melt. None of the sources of Ca melt within kiln temperatures, so they need to dissolve into an existing melt. There is a CaO-Al2O3-SiO2 eutectic at 1170C, though adding Na brings this down to about 775C. This pretty well ties in with potter's practical experience of Ca being an active flux from around 1100-1150C upwards; below this up to about 10% molar Ca can dissolve into the glaze, but it is not acting as a flux.
On cooling, particularly if there is a lot of Ca present, wollastonite crystals may form (at about 1120C), and these increase the strength of the material, particularly if they grow into the needle-like form, together with other crystals such as anorthite (CaAl2Si2O8) and gehlenite (Ca2Al2Si2O7). If there is P present, apatite crystals (Ca5(PO4)3(Cl/F/OH)) may also form, especially if there is some fluorine. The crystals are responsible for the development of matt glazes, which are less durable than non-crystalline glazes.
Once cooled, on exposure to acids the Ca will leach out, whereas alkalines will dissolve the whole glaze including the Ca.
Magnesium
The last post was on calcium, the most commonly used of the alkaline earths. So how does magnesium compare?
The biggest difference is that magnesium actually behaves as an intermediate, rather than as a flux. It can form a tetrahedra and enter the silica glaze network as a glass former, in a similar way to aluminium, or it can act as a flux in the same way as Calcium. Another example as to why we cannot see the UMF categorisations as fixed. One test series where this was found varied the alkaline earths from pure Ca to pure Mg, and when the Mg exceeded 20% of the alkaline earths some of the Mg started acting as a glass former, but destabilising the glass, reducing its melting point. However the exact conditions for this behaviour change are still being investigated.
Another paper I read that examined the solubility of Si-Al-Mg-Na glasses in a caustic solution (i.e. a dishwasher) found that it had the greatest solubility at an MgO:Na2O ratio of 0.33:0.67, which clearly refutes Matt Katz's conjecture that the maximum glaze stability is at a ratio of 0.3:0.7 (without any quantified leaching measurements, as far as I am aware). Also, other papers found that matt and crystalline glazes that use a mixture of Ca and Mg are more resistant to leaching than those just using Ca.
It doesn't dissolve into the glaze as readily as calcium. Generally it is refractory blow about 1150C.
Magnesium carbonate is the easiest to dissolve, so best used at low temperatures, then dolomite, and finally talc, which takes the most energy to enter the melt, and also gives off water at 1100C, which may give bubbles in the glaze.
Above about 0.1 molar it starts to reduce the gloss of the glaze, and from about 0.3-0.4 it strongly promotes crystallisation.
It has very low thermal expansion, and also acts as a catalyst in the production of cristobalite, both of which help reduce crazing
Beryllium (Be)
This is one material I don't have in my cupboard. Note that it is highly toxic and carcinogenic. But, for fun and for completeness, I thought I'd document its use in ceramics.
Although an alkaline earth, it acts as a glass former, for example in making beryllium crucibles - BeO has a melting point of about 2500C, it conducts heat as well as a metal, and is much more flexible than other ceramics, so withstands physical and thermal stresses much better. This means it is also used for heat sinks. But BeO is eye-wateringly expensive!
As well as BeO, it is also found in the mineral Beryl (3BeO.AlsO3.6SiO2). It acts as a catalyst for the growth of mullite, giving strong ceramics, but also cristobalite. A cone 10 porcelain may have 50% kaolin, 30% beryl, and 20% feldspar. It can also be used in glazes for low expansion, thermal shock resisting bodies such as flameware. Although emeralds are made of beryl, I haven't found anyone who can make a glaze sparkling with emeralds!
Be acts as a glass former, and always takes up the same tetrahedral structure as silica. In a Be-Si glass, it generally wedges itself into the Si network, causing some stress. For this, it needs alkali metals to give a balancing charge of +2 (which I've omitted from the drawing for clarity). As far as I can tell, it does not act as a modifier, the role of the other alkaline earths.
It seems to mute glaze colours, making them more pastel or moving them over towards grey.
In the image, we can see how a couple of Be tetrahedra have inserted themselves into a ring of six Si tetrahedra. There would also be alkali metals nearby to provide the necessary balancing charges - I've omitted them for clarity.
Alkali Earths and Crystals
In working through the alkali earths, I've been concentrating on their behaviour in a homogenous glaze, or one with phase separation. But what about crystal growth? Here I'm talking specifically about the small crystals that make a gloss glaze turn matt, rather than the large crystals associated with crystalline glazes.
People have asked me about the durability of these crystals in the past, but I hadn't seen any data on it. Observation had told me that these matt glazes were more prone to staining from the likes of tea (especially black without milk), red wine, and turmeric in curries, but I had yet to find a paper on the staining mechanism. Also, I suspected that the boundaries of crystals would be more prone to attack, as things generally are whenever a discontinuity like this occurs. But I didn't have anything on the durability of the crystals.
Until Russell Coote sent me the abstract of the paper "The impact of wollastonite and dolomite on chemical durability of matte fast-fired raw glazes" by Kronberg and Hupa, which sheds some light on things, together with the related paper "Corrosion of the crystalline phases of matte glazes in aqueous solutions". Unfortunately the full papers are locked behind an academic paywall - if anyone can send me a full copy of either paper I'd be grateful.
Anyway, summarising what can be gleaned from the abstracts, they found crystals of diopside (CaO.MgO.2SiO2), anorthite (CaO.Al2O3.2SiO2), plagioclase (a mix of albite Na.AlO2.3SiO2 and anorthite), wollastonite (CaO.SiO2), and pseudowollastonite (a high temperature phase of wollastonite).
Tests were carried out as per ISO 10545-13, which tests ceramic tiles using a range of acidic and alkaline solutions - similar to but different from tests for tableware. Diopside crystals weren't attacked by acids or alkalis. Plagioclase was only attacked in the strongest acid solution, and then just along the crystal boundaries. Anorthite and albite are attacked by mild acids if their crystals are in a plate structure, but needle-like crystals require a strong acid; both are impervious to alkalis. In an acid, the Ca in wollastonite dissolves first, followed by Si, but in an alkali they dissolve at the same rate; pseudowollastonite is attacked much more readily than wollastonite.
Crystal Formation
So we get these crystals forming in the glaze, but what are the requirements for them to form?
We need the right elements to be available in the glaze. This means not just having them present in the glaze recipe, but also not having them preferentially bound up with other elements.
The series of glazes in the image are a good example. The base recipe is 0.22 Na2O, 0.3 CaO, 0.21 MgO, 0.27 ZnO and 3.37 SiO2. Al2O3 was varied from 0.414 (glaze 1) to 1.12 (glaze 17). Glaze 1 showed extensive diopside (MgCaSi2O6) and a trace of alumina crystals; glaze 2 just a trace of diopside; glazes 3 and 4 had no crystals; and 5 and 17 had gahnite (ZnAl2O4) - just a trace in glaze 5.
Note that the areas of crystal formation do not correspond to the matt and semi-matt in Stull's paper - his glaze recipe was 0.3 K2O and 0.7 CaO, so the chemistry is quite different to this recipe, and so the positioning of the zones will also be different.
But back to the question as to why some but not all glazes grow different types of crystals.
If you remember, Al2O3 needs to pair up with balancing charges from alkali metals (preferably) or alkali earths to enter the glaze matrix. If you do the maths (or use the spreadsheet I give to my Food safe and stable glazes students), you'll see that in glaze 1 all of the alkali metals, and about 3/4 of the alkali earths have bonded with the Al, leaving 1/4 of the alkali earths free to join in with the Si and grow diopside. And as the coupling between aluminium and alkali earths is imperfect, there's a little free alumina to crystalise out. In glaze 2, the increased Al consumes just about all of the alkali metals, and only a trace of diopside was found.
In glazes 3 and 4 things are pretty well balanced out, so we get a totally homogenous glaze.
Then in 5 and 17 we have an excess of alumina in the glaze, and this forms gahnite with some of the zinc. It does this in preference to using the Zn to form a tetrahedron and enter the glaze matrix.
So basically we have two situations: first, where there is an excess of alkali earths we will get them form crystals with Si, as discussed last week. Secondly, where we have an excess of Al (due to a shortage of alkali metals and earths) we will get the Al forming crystals with Zn. So crystals will only form when we have a recipe that contains an excess of alkali earths, or an excess of Al.
Barium and Strontium
These two alkali earths have similar behaviour, though strontium is generally more refractory, so a higher temperature is needed, and sometimes the colours can differ slightly.
Many potters substitute barium with strontium, because of concerns over toxicity. Although strontium is probably not as toxic as barium, it has not been studied in as much depth. Furthermore, in moderate quantities barium glazes are very stable. The problem is with high barium glazes.
In low proportions, barium and strontium act as a standard alkali earth. They will act as a network modifier, joining silica or alumina tetrahedra with their two weak ionic bonds, weakening the glaze matrix; and they will also provide a balancing charge for the intermediates, enabling them to enter the glaze matrix as tetrahedra, if all the alkali metals have been used up.
Compared to a calcium glaze, barium makes the glaze softer and less durable, with a lower melting point. But on the plus side it has a higher density, and so refractive index, so it will be a more brilliant glaze, almost as good as a lead glaze with the same amount of lead in it. I haven't found data on this for strontium, but it is reasonable to assume that it behaves in a similar way, though with a higher melting point than the barium glaze.
In higher proportions they will form celsian crystals in the glaze - these have the formula of R Al2 Si2 O8, where R is either Ba or Sr. They can take up a number of crystalline structures, and the mix of forms is determined by the cooling of the glaze. This goes from clear crystals with a fast cool/low temperature hold, then through white crystals and back to clear as the hold time and/or temperature increases. In addition, R2 SiO4 crystals may form. Crystal growth is helped by all the normal seeds, but tin seems to be particularly effective. Note that there is a phase change at around 350C for barium, or 700C for strontium, but the transition is slow and so unlikely to cause cracking. Also, the barium crystals have a lower CTE than strontium. Note that additions of small amounts of Li (<1% molar) results in phase separation between the silica glaze and the celsian.