I've deduced that there are X number of requirements needed to get phase separation.
Silica : Alumina Ratio of 11 to 12
I think this is one of the most important requirements. The molar ratio of silica to alumina must be around 11 or 12. If the ratio is lower than 10, the glaze becomes a celadon without phase separation. If the ratio is higher than 13, the glaze loses its blue color.
Twice as much calcium oxide as alumina (molar)
This is approximate. The calcium requirements seem to be rather lax. I haven't done much testing. This is equivalent to equal weight percentages of CaO and Al2O3
Enough flux
There seems to be a requirement for NaKO ratio for phase separation. I've seen phase separation in glazes that differ in feldspar content by 20%, so the window must be rather wide.
At cone 6, there needs to be enough boron.
It seems like a thick, gloopy melt is needed for best phase separation. Too stiff and the melt can't phase separate, resulting in a muddier color
~1% Fe2O3
Approximately 1% iron is needed to develop the blue color. Without iron, the phase separation is colorless.
Version 1: Initial Test
Existing cone 10 chun recipes have application issues due to the fact that there is almost no clay content. By using Redart, a high iron terracotta, my glaze will have better suspension and dipping while also saving me from measuring small amounts of iron for test glazes.
For the first test, I started with 20% boron frit and 30% redart. The remaining ingredients are feldspar, silica, and wollastonite. I tuned the percentages to approximate the ratios in Wiliam Chau's analysis of jun glazes. I also aimed for a 0.3 R2O to 0.7 RO flux ratio for durability.
I varied the amount of silica and it seems like the increased silica improves the phase separation. The version with only 10% silica strangely came out matte.
Note that the medium and high silica recipes total to 105 and 110 respectively.
Version 2: Lower Redart Content
Because the glaze were more green than blue, I decided to lower the amount of Redart (and therefore iron). I kept the boron content the same and targeted similar silica, alumina, and calcium ratios.
Because the 10% silica test tile from the first run came out matte, I was worried that my Redart's silica content differed from online analyses. To empirically verify the amount of silica needed, I again kept the alumina and calcium content the same and varied the silica content.
Material
Low Silica
Medium Silica
High Silica
Custer Feldspar
30
30
30
Whiting
10
10
10
Ferro Frit 3134
20
20
20
Silica
20
25
30
EP Kaolin
5
Redart
15
15
15
Note that the high silica version totals to 105 instead of 100%.
This firing went hot, reaching cone 7 or thereabouts.
Version 3: Altering the sodium / potassium content
This test gave negative results. None of the test tiles had strong phase separation.
Because of this test, I believe that William Chau's phase diagram analysis is misguided. The phase separation is not a single point in the silica-alumina-calcium phase diagram with chemistries closes to the point having the best color. I think the silica-alumina ratio is far more important than the calcium content.
In all of these tests, the phase separation sharply dropped off when the silica-alumina ratio fell below 11. However, there was still significant separation above that line.
Version 4: Full Triaxial
This triaxial had disappointing results. There was no consistent pattern of phase separation. Even when there were traces of phase separation, it was weak compared to previous tests.
Version 5: Altering the flux content
Version 6: Altering the boron content
Initial test
Observations
Existing glazes can get good results without bone ash (phosphorus for phase separation)
William Chau says that the strength of the blue color is primarily caused by the ratio of alumina to silica to calcium oxide
Existing glazes have poor suspension qualities due to low clay content
I don't like measuring small amounts of iron for glaze tests
Little to no recipes online at cone 6
Hypothesis
I can solve the suspension and iron measuring issues by using Redart (commonly available American terracotta clay)
I can lower the glaze melting point from cone 10 with the addition of boron frit (0.25 molar B2O3)
If I alter the quantity of silica, I should get stronger and weaker phase separation
Experiment
Created glazes from Redart, FF 3134, Minspar, Silica, Whiting
Altered the silica percentage by adding in excess of 100%
Results
There was slight phase separation in the 10% and 20% extra silica tests
No phase separation in the chemical match to Chau's theoretical Jun
Conclusions
I'm on the right track because I was able to get some phase separation. The only issue now is increasing the strength
Maybe my Redart is silica deficient compared to online chemical analysis
Second test
Observations
Test 1 tests were green where phase separation is weak
Excess of silica still produced phase separation
Hypothesis
I can reduce the green color by reducing the Redart content (iron)
My actual materials differ from what the glaze calculator says
I can create Jun
Experiment
Reduce the Redart content from 30% to 15%
Alter the silica to alumina ratio by increasing silica or adding small amounts of kaolin
Results
Strongest blue color in test tile with lowest silica to alumina content
Some reds and purples on the edges of test tiles
Kiln over fired to cone 7 to 8
Third test
Observations
Unwanted bubbles in second test
Strongest blue had surprisingly low silica content
Some purples and reds on test tiles, perhaps from flashing copper in nearby pots
Hypothesis
I can reduce the bubbles by replacing whiting (calcium carbonate) with wollastonite (calcium silicate) and adjusting the silica content
Since I have a glaze that works, I can check the "region around it" by creating a triaxial with differing alumina, silica, and calcium content
I can create red colors by adding copper (excavated Jun glazes have copper splashes)
Experiment
Create three new recipes based on successful one from test 2: one with too much alumina, too much silica, and one with too much calcium
Systematically blend the three glazes to create a three way gradient
Partially spray test tiles with a weak copper sulfate solution (1wt%)
Results
No phase separation in high alumina tests
Some phase separation in low alumina tests
Best reds and purples when copper is sprayed thin on low alumina tiles
Fourth test
Observations
Best jun glazes are slightly drippy
Tiles that were overfired in the 2nd test had better phase separation than any tile in the 3rd test
Hypothesis
If I refire the low silica glaze from test 2, I will get less phase separation at cone 6
If I dissolve copper sulfate into the glaze mix, I'll get more even and brighter reds
If I substitute wollastonite, I won't have any changes in phase separation
If I increase the feldspar content, I'll get an increase in phase separation and glaze movement
Phase separation is not dependent on three way relationship between silica, alumina, and calcium, but two independent factors: 1) a silica and alumina ratio of 11 and above and 2) some quantity of calcium
Experiments
Results
Better phase separation than 4th test, but strength is not correlated to feldspar content (bar measuring errors, but I've been able to mix glazes very consistently in the past)
Refiring v2 low silica did not produce strong phase separation
Fifth Test
Hypothesis
Increasing the boron content can provide sufficient melt flow for phase separation
I think this is the 7th test? Lowering alumina content may be a good idea. It seems like all blue jun glazes have some movement. They're all a bit gloopy. Silica-alumina ratio is preserved, as well as calcium ratio (extra flux is derived from talc in the form of magnesium). 7/16/24