Predicted COE (Coefficient of Expansion) - trust your actual results rather than the predicted COE

[Norm Stuart]

One of the helpful tips Tony Hansen of Digitalfire has passed on to me is the "calculated thermal coefficient of expansion" in Insight-Live and other glaze software is not very accurate when used to predict the COE of a clay body.

We know the COE predictions for clay are often wrong because it's very easy to measure the COE of a fired clay in a dilatometer, shown below as a photo and a diagram.

Our supplier, Laguna Clay, provides the actual tested COE and other specs for most of their clays. - http://www.lagunaclay.com/clays/western/wc861.php

Actual tested COE was an important piece of information when making a "flameware" clay body requiring very low expansion because small changes, like tap water instead of distilled water, make significant and undesirable changes in the actual COE of the fired clay body. These large changes in the COE of the flameware clay body are not at all predicted by Insight and other glaze software.

This always left me curious why the predicted COE for clay would often be inaccurate, but the predicted COE for glaze would be correct.

I've come to realize that the same math models are also probably not always accurate when predicting the C.O.E. of glaze as well.  Definitely something to bear in mind.

One example is Boron which is both a flux, and a glass former like silica. Ceramic research papers show can allow a glaze to fit clay bodies with a much wider range of COE than you might expect by creating a variety of different structures depending on the force and movement while the glaze is cooling.  Boron can also expand the temperature range of structural stability for vitrified porcelain.

Digitalfire - Adjusting a Cone 10 recipe to Cone 6

http://www.borax.com/docs/euf_pdfs/euf-borates-boratesinglazesanden...

A practical example of this is Zam Celadon ^6, a glaze fluxed primarily with Nepheline Syenite which contains 9.8% Na2O and 4.6% K2O creating a relatively high COE of 7.4, very often resulting in crazing.

This is terrific if you want to achieve this crackle look of a crazed glaze which makes it unsuitable for functional use.  This glaze will even crackle when applied to Laguna Clay Frost WC-437 which has a measured body COE of 6.99 which is fairly high for clay.  Most would think a glaze with a COE of 7.4 over a clay with a COE of 6.99 should be a good fit, but not in practice.

This glaze contains Ferro Frit 3124, which is relatively low in Boron Oxide (B2O3) - only 13.74%.

If I replace the Ferro Frit 3124 with Ferro Frit 3195 (a balanced low-fire clear having 22.62% Boron Oxide), the percentage of Boron Oxide B2O3 increases by 82% from 0.17 to 0.31 but very little else of the glaze chemistry is changed.

The predicted COE declines a tiny amount from 7.4 to 7.2 but all crazing ceases regardless of which clay body I've used it on.  So the extra Boron Oxide provides additional fitting ability which is not captured in the predicted COE.

One drawback is a glaze with a high percentage of early melters, such a high boron oxide percentage, can be counter-productive if there are glaze ingredients which continue to off-gas above Cone 2, as the boron will likely have already melted the glaze by this temperature, which would lead to the glaze foaming, leaving it with pinholes if the melt is too viscous.

Pete Pinnell has suggested that if we had relied exclusively on Glaze Limit Formulas we would be missing most of the interesting glazes which have been created over the last 50 years, and he a makes a pretty compelling case for that.

Pete's Weathered Bronze Green, which can be used at Cone 10 or Cone 6, with a predicted COE of 8.7 is just one of many examples. With a Silica to Alumina ratio of only 4.7 to 1, Weathered Bronze Green is more of an engobe rather than a glassy glaze.

Sometimes our biggest problems comes from the false things we "know" to be true rather than things we're pretty sure we don't know.

If you experience crazing or shivering, comparing the COE or the clay with the predicted COE of the glaze can help give you an idea of how to fix the problem, but test tiles will tell the real result of your adjustments rather than the predicted COE.  Added Boron is just one of the ways to fix a COE mismatch.  Cooling glaze more slowly can at least give the appearance that mismatch problems have been cured as no glaze defects will be visible after firing.

The logic used to predict COE is trying to accurately model often very complex ceramic structures

As ceramic researchers learn more, they way COE is predicted will almost certainly become more accurate as we learn to better model the real world. Currently test tiles are our best friend.

[Marina Reijsmeijer (Kleierij)]

Interesting. It confirms resutls of my own tests where I have glazes with COE 5.4 to 10.6 that do not craze. The 10.6 one does not have Boron in it either.