Applying a very thin layer of clear glaze over painted porcelain

[thiamant]

21 minutes ago, Min said:

edit: how far apart are your craze lines

 

And with the zinc:

 

[Min]

I don't think the simple fix of adding silica will fix the top glaze. It's hard to clearly see how much crazing is in the second one with zinc. 

Do you have any Spodumene or Petalite? Lithium carb? Are you getting to a full cone 10? Checking with witness cones?

[Mark C.]

You could just use the recipe I posted as it works well at cone 10

[thiamant]

59 minutes ago, Min said:

I don't think the simple fix of adding silica will fix the top glaze. It's hard to clearly see how much crazing is in the second one with zinc. 

Do you have any Spodumene or Petalite? Lithium carb? Are you getting to a full cone 10? Checking with witness cones?

Amazingly all the g1947u came out perfect with 0 crazing and passed the stress test...

I have spodumene. I guess i could add in a bit of lithium to get to a .2:.8 flux ratio since im slightly below the .2 (just like the g1947u). In fact just adding 2% spodumene i get

Is that going to make a big difference?

Why is the g1947u not crazing at all? Is it the zinc? The potash feldspar?

@Mark C. I dont have ball clay in my studio but crazing is more body dependent so your glaze might work perfect for your clay body and crazy crazing in my clay body...

 

 

[Min]

51 minutes ago, Mark C. said:

You could just use the recipe I posted as it works well at cone 10

 

3 minutes ago, thiamant said:

I dont have ball clay in my studio but crazing is more body dependent so your glaze might work perfect for your clay body and crazy crazing in my clay body...

 

If you remove the zinc from Marks recipe it is virtually the same formula (molecular) as the 1947U. Screenshot below. Wollastonite to supply the calcium (in place of whiting) and ball instead of kaolin. Makes me think that either might work if it was fired hotter. I'm guessing the 1947 was based on the same recipe Mark uses.

5 minutes ago, thiamant said:

Why is the g1947u not crazing at all? Is it the zinc? The potash feldspar?

Likely because it's lower in sodium. I've got to get some work done here but I can look at your spod version later today if someone else doesn't do it before me. If you  post the spodumene recipe you are thinking about it will be faster. It's good to know you aren't getting crazing with the g1947u. Are you verifying you are getting to cone 10 with witness cones? Underfiring could account for both the crazing and the cloudy look.

 

[thiamant]

@Min

Ok, so this is my version of the g1947U with grolleg kaolin:

https://glazy.org/recipes/163310 (not crazing)

This is the NephSy clear with 15% kaolin (also tried with 18% with same results):

https://glazy.org/recipes/163307 (crazing)

And the zinc version of the nephsy clear is just the previous one with 18% kaolin and 2.5 zinc added.

And the spodumene is just https://glazy.org/recipes/164474 It's very little spodumene maybe I should just add more and remove some nephsy while keeping ratios.

 

15 minutes ago, Min said:

Are you verifying you are getting to cone 10 with witness cones? Underfiring could account for both the crazing and the cloudy look

Yes, These tiles were at the bottom of the kiln next to witness cones, the lower part of the kiln fires a bit lower, so cone 10 went to 3 o'clock, cone 9 went completely melted and 11 started bending.  (The upper part of the kiln fires to a perfect cone 10)

[Mark C.]

My clay  currently is porcelain and as you noted testing different bodies will answer all the questions . I have used that clear on many white clay bodies over 40 years with great results.Many of my fellow local potters used it as well. No ball clay in Spain ?must be substitute .

[Min]

I had a quick look at your Feldespato potásico extra that you used in place of Custer feldspar, there are enough differences in the sodium and potassium that just might make a difference in the melt. Your Nefelina Sienita Prodesco also is a bit different, both mostly in the sodium and potassium levels. I'ld do a side by side comparison of both recipes using your ingredients compared to North American. Custer Feldspar and Nepheline Syenite A270 and look for differences.

Other thing I noticed is you didn't rebalance the calcium in your Nefelina Sienita Prodesco version. Screenshot from Glazy of you Grolleg + feldspar on left compared to the Nefelina Sienita Prodesco. Calcium is one of the higher expansion oxides, dropping that down will reduce crazing but I'm not sure it's enough to eliminate it.

If you go with the Neph Sy version (using your Spanish material) I'ld go for about 1/2 that and 1/2 spodumene to bring the expansion down. It will probably put the flux ratio in the 3:7 range but should still be a durable glaze.

I'ld also try the original or grolleg version again and just thin it down some more. Easiest fix would be if a thinner layer is all it takes. Take the sg down to about 1.3 and just do a quick dip, see if that does it.

Might find the same problems using Marks version of this glaze since it's pretty much the same glaze but with a higher loss on ignition and a tiny bit more iron.

 

 

 

 

 

[thiamant]

@MinI compensated the lack of Zinc with more calcium to keep fluxing ratios more or less the same... I never considered adding more calcium would make the glaze craze more. I don't trust the COE given by calculators since they are based in raw materials and not the actual melt that you end up getting, which is quite different. 

[PeterH]

33 minutes ago, thiamant said:

I don't trust the COE given by calculators since they are based in raw materials and not the actual melt that you end up getting, which is quite different. 

Are you sure that you aren't overstating your case? Especially for the relative expansion of   homogeneous glazes; e.g. fully melted glossy glazes?

As suggested by Ron Roy, chapter 5 of Mastering Cone 6 Glazes makes interesting reading here.

See expansion of clay vs expansion of glaze http://www.potters.org/subject58970.htm
... which uses glazes of different expansions to discover the expansion of a clay body (as fired to a specific schedule).

As you can see in chapter 5 of Master cone 6 glazes - using glossy glazes
to determine body expansion will work - and you can design the glazes using
calculation software - there are some stumbling blocks to watch out for -
don't use Lithium Carb and keep the Boron below 12% - and make sure the
glaze is properly melted - but it is not rocket science.

 

[thiamant]

35 minutes ago, PeterH said:

Are you sure that you aren't overstating your case? Especially for the relative expansion of   homogeneous glazes; e.g. fully melted glossy glazes?

As suggested by Ron Roy, chapter 5 of Mastering Cone 6 Glazes makes interesting reading here.

See expansion of clay vs expansion of glaze http://www.potters.org/subject58970.htm
... which uses glazes of different expansions to discover the expansion of a clay body (as fired to a specific schedule).

As you can see in chapter 5 of Master cone 6 glazes - using glossy glazes
to determine body expansion will work - and you can design the glazes using
calculation software - there are some stumbling blocks to watch out for -
don't use Lithium Carb and keep the Boron below 12% - and make sure the
glaze is properly melted - but it is not rocket science.

 

Unmelted silica has super high thermal expansion, while melted silica is the opposite, super low thermal expansion. How am I supposed to trust this system which is based on unmelted materials?

[Bill Kielb]

1 hour ago, thiamant said:

Unmelted silica has super high thermal expansion, while melted silica is the opposite, super low thermal expansion. How am I supposed to trust this system which is based on unmelted materials?

My understanding, calculated is not a great representation of fired so it’s use as more of a trend or a relative comparative value than a precise number is likely the argument. I feel your pain, I try not to get trapped into designing to a number or range and believe for me it’s easy to hyper focus on the appearance of numeric precision. From a trend standpoint though it seems to have value. IMO

[Min]

3 hours ago, thiamant said:

Unmelted silica has super high thermal expansion, while melted silica is the opposite, super low thermal expansion. How am I supposed to trust this system which is based on unmelted materials?

You just pointed out one of the parameters of using (reversible) COE or CTE figures. They do have their limitations, first off the glaze mustn't be crystalline; it must be fully melted.  Other factors that need to be taken into consideration are glaze elasticity, oxides that don't move in a linear fashion such as lithium and boron  and while comparing formulas the same oxides need to be present in the ones being compared. If you want to test this increase the Nefelina Sienita Prodesco content of your glaze and see if the crazing doesn't get worse. (while balancing the silica and alumina)

A simple and sometimes effective way to solve crazing is to add silica and alumina, this will work if and only if the glaze will melt enough to incorporate the additions. Not to keep banging on about it but the glaze must be a fully melted gloss glaze to utilize COE figures. However by understanding the properties of the oxides we can utilize the low versus high or somewhere in between information for claybodies and matte glazes also. Example of this is the stovetop cookware that utilizes spodumene or petalite. Plunk the figures into a glaze calc program for one of these claybodies and it will give you a COE figure but it's not applicable since the clay is a crystalline structure not a fully melted homogenous matrix and yet we know a flux containing lithia is often used in  very low expansion stovetop claybodies.

COE is measuring the reversible volume of change of a ceramic oxide when heated and cooled back down. This is relevant since that is what our glazes undergo. Using measured COE figures is another tool in the tool box and like all tools needs to be used appropriately.

We know what the COE of the oxides we use are (given they could be slightly different depending on source). On the high end of the expansion oxides we are looking at sodium and potassium and on the low end we are looking at lithium, magnesium, silica and alumina. 

@BobMagnuson has permitted this forum to have access to his Eutectics Calculator. I entered the original G1947U glaze into it. Very slight oversupply of slica (1.43) and alumina (0.02). If you are questioning an oversupply of an oxide it's a great tool to utilize. (link at the top of this section)

[PeterH]

Sorry I'm not trying to start either a discussion or an argument. I just felt the fact that some/many experienced potters seem to value these predictions as an aid to tailoring some glazes might be a reason to reconsider your viewpoint. [ Especially for the relative expansion of   homogeneous glazes; e.g. fully melted glossy glazes.]

1 hour ago, thiamant said:

Unmelted silica has super high thermal expansion, while melted silica is the opposite, super low thermal expansion. How am I supposed to trust this system which is based on unmelted materials?

The interest in COE is to avoid crazing, a property of solidified glazes. So a system based on the properties of unmelted materials seems a natural choice.

[Min]

3 hours ago, PeterH said:

The interest in COE is to avoid crazing, a property of solidified glazes. So a system based on the properties of unmelted materials seems a natural choice.

Plus to avoid the opposite, shivering.

I believe there are 3 main methods of measuring an oxides COE / CTE, they all measure a sample during a thermal cycle. Ceramic oxides, like most substances, expand when heated and contract when cooled, this is what is being measured.

 

[Hulk]

I'm seeing that the calculated coe values don't necessarily line up when comparing very different glazes and their behaviors on the same clay bodies, however, when looking to tweak a glaze (to limit/eliminate crazing - haven't had the joy of shivering yet), then I'm finding the calculated coe very helpful, where I've lowered two high expansion oxides, increased low expansion oxides, and bumped up the boron a bit get a lower calculated coe, and less crazing ...then more changes, lower coe, even less crazing, more adjustments, aha! fitted glaze.

A bit of lithium can make quite a difference; I'm using petalite, was looking to avoid the foaming, and it was inexpensive as well. 

Magnesium (via talc), also a handy low expansion oxide...

[thiamant]

I'm going to try 0.15 LiO2 + 0.15 KNaO2 instead of 0.3 KNaO2 and see what happens. 

I haven't had these foaming problems yet (only used spodumene in small scale).

[Min]

Might want to try lowering the KNaO2 even more. Along the lines of 0.15 LiO2 and 0.11 or 0.10 KNaO2. Since you are firing tests anyways do a few of them to save firing more than you need to. Lithia could be bumped up to 0.20 I would also suggest dipping / spraying different thicknesses of glaze, might be able to really thin it down which will likely make a difference with the clarity of the glazes.

I'ld also add your materials into the Eutectics Calculator and see if the glazes are oversupplied in some of the fluxes. That calculator is new here, if you can't enter your spar and nepheline syenite you could DM @BobMagnuson and ask him about it.

 

[thiamant]

So here are the results after thermal shock stress test, I tried to keep my ratios .3:.7 (.15 Lithium .7 potassium .8 sodium) and tried with different silica levels, decreasing silica levels really makes things worse ( I was hoping maybe less cloudiness by doing so, but all tests from 1 to 4 had crazing). Test 5 didn't show crazing on body 1 but it did on body 2.  Test 9 had the most Silica and alumina, and didn't craze on any of these bodies. Test 8 is the only one that didn't have .3:.7 because it's just g1947U minus the zinc. So it actually has .17:.83 flux ratio. Looks melted but crazing a bit on body 2 (maybe because of unmelted silica). Test 6 is lying really close to the original g1947U but in this case it is crazing (unlike the original g1947u with the zinc), probably because there are more alkali metals fluxes in it.

#9 is on the limit of the melting area (.58 Alumina 4.63 Silica), but it shows the best results. Many of the applications were on the thin side so I will repeat 9 a bit thicker and probably another one with less amount of silica. I could also try more lithium and less sodium, or even .2:.8 with .2 Lithium. Since I'm already really high on the si/al levels, that also means less spodumene and cheaper glaze... I'll see depending on how much time I got...

 

EDIT: Sorry tiles 3 and 4 are interchanged, they do follow the more silica less crazing trend (I placed them wrong in the picture).

[Bill Kielb]

52 minutes ago, thiamant said:

#9 is on the limit of the melting area (.58 Alumina 4.63 Silica), but it shows the best results

Just curious,, these are fired to cone 10 as I recall, so what limits are you applying? Anyway you can publish the COE of each recipe?

[thiamant]

15 minutes ago, Bill Kielb said:

Just curious,, these are fired to cone 10 as I recall, so what limits are you applying? Anyway you can publish the COE of each recipe?

I don't use limits, just Stull, UMF and chemistry, but they are all within green & cooper and Hesselberth & Roy cone 10 limits. Only 6 and 8 are within Val Cushing's cone 9-10 glossy limits.

The calculated COE are:

1 - 6.79

2 - 6.62

3 - 6.47

4 - 6.33

5 - 6.20

6 - 6.32

7 -6.27

8 - 6.48

9 - 5.99

[Bill Kielb]

2 hours ago, thiamant said:

#9 is on the limit of the melting area (.58 Alumina 4.63 Silica), 

 

1 hour ago, thiamant said:

don't use limits, just Stull, UMF and chemistry

Interesting  COE range!  What is the limit of the melting area?

[thiamant]

23 minutes ago, Bill Kielb said:

 

Interesting  COE range!  What is the limit of the melting area?

From my experience and the real experiments I have seen from different people, usually more than 0.6 alumina and 5.5 silica is going to have a hard time melting. I dont like limits because they Just put a rectangle in there and its not that simple. In reality its more like a curve with different gradations of melting... But thats probably another discussion for another day... I am curious why did you ask? 

What I wanted to say is that test #9 is probably far from the ideal melt at cone 10, it looks a bit dry but since the application was thin I cant really tell, so I have to repeat this one and probably a couple more with a thicker application to really find the best candidate. 

[Bill Kielb]

6 minutes ago, thiamant said:

But thats probably another discussion for another day... I am curious why did you ask?

For cone ten stuff those numbers would not bother me especially if it melted nicely so I was just curious where they came from.

[Bill Kielb]

4 hours ago, thiamant said:

The calculated COE are:

1 - 6.79

2 - 6.62

3 - 6.47

4 - 6.33

5 - 6.20

6 - 6.32

7 -6.27

8 - 6.48

9 - 5.99

It is interesting, If I look up the calculated COE of the original glaze recipe it is 6.5 which puts it pretty much in the middle of your tested  range and it did not craze.  I also see several references to successful use along with pictures that appear reasonably clear. I wonder if something is being overlooked here.