Saturday, December 25, 2010

Testing and developing glazes

I've been gearing up for some serious glaze testing. I've been watching the struggles of my friend, Kim, who is developing slips and glazes for soda firing. 

Building on multiple sources, I have a plan for how to proceed with my own testing. In this blog, I will share my ideas and any good glazes I develop.  Comments or suggestions are welcome.

Why is testing necessary?

Glaze recipes do not travel well - often need to be adjusted for local conditions. Making a single test tile of a recipe is a shortcut that often misses the mark.  Likewise, trying to use glaze calculation software to create a one-shot cure is also a shortcut that often misses the mark.

What sort of local conditions can affect glaze features?  Hot oxides interact with each other, the kiln atmosphere, and the clay body ...  and have complex effects on the glaze melt.  There is a lot of variability in raw materials, clay bodies, solubles in the water, kiln atmosphere, and firing schedules - any of which can affect the glaze.  Glaze calculation is no more accurate than the analyses of the materials - should be taken as a rough approximation.  Your working recipe will be the result of systematic testing.  Use the software to narrow the scope of the testing.

When to test

Test all new glazes and test working glazes any time something changes, such as a change in raw material, glaze density, water supply, clay body, firing schedule, kiln atmosphere, colorant combinations.  You also may need to test for suitability for leather-hard, bone-dry, or bisque application. 

How to test

Instead of testing a zillion recipes to find a few that work for you, your time will be better spent if you select a few glazes and test them systematically ... on the faith that you can probably get any of them to work.

You will want to do either a line blend (varying one oxide) or a square blend (varying two oxides). Rather than weighing out all the ingredients of each test, you weigh out just the ingredients for each boundary point of the blend (such as the corner glazes of a square blend or the ends of a line blend), add water to make them have equal volumes, then use volumetric means to proportion out the boundary mixes to create the mix for each test tile.  

After the tiles are fired, glaze software can instantly provide the recipe for every tile in the blend based on the recipes of the corner glazes and the number of intermediate points. 

The commonest variables that require testing are alumina and silica - something that should be done with any untried or problematic recipe.

You have to keep things simple in order to make sense of the results.  In every test (even a square grid that changes two variables) you are really always examining a change in single variable at a time while you keep everything else the same.  

Currie blend (or grid) is a type of square blend that fixes the proportions of a set of glaze ingredients while systematically varying the alumina and silica to create a family of different glazes.  This grid is usually 5x7 (35 tiles).  Silica increases by 4 increments left-to-right.  Alumina increases by 6 increments bottom-to-top.  Using the same size grid helps you visualize a tile's position in the grid and understand how the glaze changes as you vary alumina and silica.

A more general concept is the square blend in which you systematically vary one oxide in each of two axes.  You create corner recipes that determine the oxide levels in the corners.  You determine the number of increments in each axis.  You can customize the scope and granularity of the grid any way you want.

I believe that the best use of glaze software in solving glaze problems or in developing new glazes is in the design of tests.  Reality is what is recorded in the tiles.  Your working recipes will be the recipes of the mixes you actually put on the tiles - not something you came up with using software to juggle ingredients while you check the calculated effects on oxide levels.  

You can design the corner glazes of your test to make the range of oxide levels being tested as broad or narrow as you wish.  You can determine the granularity of your testing by defining the number of intermediate points in each axis of the test.  

If you are trying to fine tune an almost-working recipe you want to narrow the focus of your test.  If you are trying to discover a glaze nobody that has never seen before, you can make the scope as wide as you want.  If you are a functional potter, you can use the software to keep the oxides in the four corners close to the known limits for durable glazes.

In constructing a square grid, you define a set of ingredients that will be constant in all 4 corner glazes.  Then you add ingredients to the corner glazes that will systematically change only one variable in each axis.

Labels, registry, computer records

Each test set and each tile must have unique identifiers. The computer can be used to define the tests before they are run ... as well as the labels that identify each test and it's corresponding set of tiles. There should also be a hand-written registry for recording date, test, and tiles.

The hand-written registry will only show what tests were done and when. The details are all in the computer in records that can easily be backed up. If the registry book were lost, it would be no big deal because the information originated in the computer records. (If the computer records were to be lost, the registry information would be nearly useless.)

The test set identifier will consist of a letter corresponding to the year plus a sequential number representing the number of the test set during that year. For example, if "A" represents "2010", then "A46" would be the 46th set done in 2010, "B19" would be the 19th set done in 2011, etc.

There are difficulties involved in labelling tiles being tested in salt/soda firings. Identifiers painted on the bottoms of these tiles are easily obscured by kiln gunk. One solution to this problem is to stamp or inscribe the identifier on the tiles during the leather-hard stage. Another way would be to paint the identifier onto an unglazed area on the back or side of the tile, rather than on the bottom.

The most precise and error-free way to identify the tiles in a set would be to paint a label on each tile immediately after it is glazed, using a combination of set identifier plus the sequential number of that tile in that set. For example "A46-21" would identify the 21st tile in the in the 46th test set of 2010.

Glaze software

Glaze software is indispensable for understanding and developing glazes, and for keeping records.

The different available glaze software products vary considerably in their ability to store and organize information. GlazeChem is a good one for storing and retrieving records.

The records in GlazeChem can include images of glazes and detailed notes.  You can keep all your tests in special folder.  Each test can be kept in a separate glaze datebase file on the computer. Each database can contain detailed notes, the recipes of the boundaries of a blend, and the recipes of all the tests in the set. It is possible to predefine and label each test set and print out the corner recipes before you go out to the studio to mix the glazes. The printouts can be used in the studio to check off glaze ingredients as they are added to a batch. After the tests are run, the print-out recipe lists can be discarded. You can use the program to determine the recipe of any of the tiles that interest you.  You can go back to the computer and record your results ... even including images of selected tiles.  You can give descriptive names to tiles/recipes that worked well and copy them to a collection of working glazes.  

Thursday, December 16, 2010

Hard to please #1

A recent sculptural vessel I'm calling "Hard to Please #1"...  The height is about 20 inches.  Stoneware.