search  current discussion  categories  glazes - misc 

winkelmann and schott coe figures and an insight question on coe

updated sun 22 jul 07

 

Dan Semler on thu 19 jul 07


Hi Karen,

A quick response before I have more time to digest all that you have said=
.

1. I agree that the boron thing is problematic given the behaviour =20
of 3134/F12. Thought of that at the time, but figured better to toss =20
it out there.

2. COE figures are calculated from test series of glazes that use a =20
number of glazes of various compositions and then measure their =20
expansions. They then plot or calculate the contribution of various =20
oxides to the expansion. David Hewitt had a really good paper on his =20
website on this comparing a number of the key sets of values. Not all =20
researchers calculated values for the same oxides, alas. And of =20
course, they don't always the same results :) I don't know what values =20
Insight uses for its default COE set. So there could be something =20
there. I had a problem with my insight MDT until night before last. I =20
will look further into this and see what's up. It certainly makes =20
sense that the COE calculations could be at fault in this confusion =20
given the evidence. Your method sounds fairly well controlled so =20
that's good. Can you let me know which MDT you are using. I generally =20
use Ron Roy's and just downloaded the latest from Digitalfire the =20
other night.

Since David pass away his site is offline, but the content is =20
available in the Web Archive at : =20
http://web.archive.org/web/20061109202618/http://www.dhpot.demon.co.uk/. Loo=
k =20
for the article on Calculating Crazing.

Thx
D

Karen Latorre on thu 19 jul 07


Dan, thanks for the response and thoughts. I'll address or comment on a few
of the items below and add some additional information that further causes
me to think the W&S COE values are off (see the end for this)

> Posted by: "Dan Semler" the point was that boron oxide's contribution
>to expansion is said to change above 12%, increasing expansion where it
>had previously it had lowered it.
Boron content: 3134 and F12 have 23.1 and 23.8 respectively, and 3110 and
F75 have 2.6 and 2.5 respectively, so I'd expect the 3134/F12 to react the
same way, as would the 3110/F75. So, that's not it.

>I've never tried a polarizing filter to look at crazing.
I don't think you could do this for crazing on a pottery body. One of the
filters goes behind and the other in front of the piece of glass and when
held up to the light you can see a "squared halo" around those areas with
internal stress, and the brighter and larger the halo, the more stress in
the material. In all tests I've done, those with the most visible halos have
cracked the glass within days or weeks.

>wonder to what extent each frit may have bonded with the window glass.
>(amount, thickness,
temp etc.,)
All frits were loaded with the same small forming tool (same quantity), the
water was measured with a syringe (same amount of frit per amount), and all
tests were run on the same piece of glass (same firing conditions). I don't
think this is a factor.

>I am wondering if you will see a different stress line pattern for glazes
>that are in considerable compression (ie. close to shivering) as you would
>for those that are crazing. I'm also not clear on the extent to which
>there is an interface layer forming between the glasses.
The glasses are melting together at the interface. What was an edge before
firing is a gradually dissappearing opacity with the highest opacity in the
center of each "frit drop". The stress lines are always the same in my
experience, and interestingly display the same whether the incompatible
glass/frit is introduced in a circular or square area. I don't know why this
is.

> Another test worth trying perhaps would be to glaze some bisqued tiles
>with the frits and see which show crazing and which don't, and see what
>light that might shed on this.
Good idea! I'll try this out at three different temps (glass fusing temp
(approx 800C/1600F), bisque (cone 06) and one of my usual glaze temp (cone 6
or 8, whichever is going through next).

I've done some further work with the INSIGHT program, and have found the
following:

Using the INSIGHT values for COE I get the following list sorted in order of
increasing COE value, as well as the following extent of internal stress
based on the two different sheets/types of glass I'm testing on (I've used a
random number to indicate amount of stress I see with 0 being no stress and
6 being cracked glass):
FRIT COE Glass1 Glass2
F2 6.9 4 6
3124 7.9 3 5
F12 8.18 0 4
F75 9.3 3 1
3134 9.6 3 2
3110 10.1 6 5

When I use the W&S numbers, the order of expanding COE values shifts with
the two Fusion Frits F12 and F75. Here's that list:
FRIT COE
F2 74.46
3124 78.08
3134 81.2
F12 81.95
3110 92.19
F75 96.61

If I use the INSIGHT COE values then it seems to match what I'm seeing in
terms of stress in the glass, but the W&S numbers don't. The only way I
could explain this is if the calculated COE values are not calculated in
INSIGHT, but are hard coded, but I've added additional materials into the
standard data set and see the COE being calculated. If it is calculated,
then it doesn't make sense that the fusion frit values would be the only
ones impacted.

I know this topic may seem "off" for this list, but I'm guessing that when
we "double or triple dip" a pot into different glazes, we're introducing
some of this interaction into our glaze layer. I'm also keeping this on list
as I know there are a number of knowledgable glaze folks here that can help
me understand the discrepancy in what I'm seeing/calculating.

There's a document on glass compatibility on one of the glass supplier lists
which may be of interest. Here's the link:
http://www.bullseyeglass.com/pdf/technotes_tipsheets/TechNotes_03.pdf

It indicates that not only is COE at play in forming a stress free glass
boundary, but viscosity, as well as manufacturing method of the glass/frit.
In looking at the chemistry of the 4 frits in question here, the amount of
Alumina in each set of 2 is quite close if not the same (0/0.8 for the
3134/F12, and 3.7/3.7 for the 3110/F75), so I doubt that viscosity is a
factor. The melting point may be at play in the 3110/F75 set though with
3110 melting at 760C and F75 at 843C. I've been fusing at 800C. The melting
point of the other set is 788C and 790C, so it doesn't explain it for that
set.

So, in short, I'm still confused and still think there's something up with
the W&S COE values I'm coming up with for the fusion frits.

Sorry for the long post, and again thanks in advance for any insight into
this,
Karen
www.karenlatorre.com

_________________________________________________________________
Windows Live Hotmail. Even hotter than before. Get a better look now.
www.newhotmail.ca?icid=WLHMENCA148

Karen Latorre on fri 20 jul 07


Edouard, thanks for the links. Interesting information.

A question on the information at the links you provided: Where I see the
term "Bass" on the graphs Edouard, this is your method of measurement,
correct?

Regarding the information at the two links, the following information is
interesting: "It should be remembered that these expansions (dilatations)
are valid only if the « Custer added material » mixture is completely melted
(or transformed into a frit). This is not valid for a paste where there
would remain nonmolten crystalline materials and badly amalgamated in the
vitreous phase."

The tests I'm conducting are using a base of 100% frit from two different
suppliers. The two frits that are in question are from the same supplier. I
would expect that the frits we use in glaze formulation have been fully
melted as part of the fritting process. Is this a correct assumption, or is
there a possibility that the frits from this one supplier have not achieved
a true full melt and therefore have some un-amalgamated portions in the
material (and thereby the calculated COE values are not matching reality)?
While I don't think this is what's causing the discrepancy (if it were, I'd
expect the INSIGHT COE values to also be off, but they're not), reading your
information caused me to wonder about this point.

The other interesting information is that W&S values are only valid from
20 to 100C whereas others are valid from 20 to 400C. Does this mean that the
W&S values don't extend well into the higher temperature levels and
therefore using them to guess at the compatibility of a glass to glass
interface would lead to incorrect values? This could be the source of the
discrepancy I'm seeing between the W&S values and the INSIGHT values of COE.

Is there information somewhere on the processes used to measure the
various sets of COE data? I don't think this is a factor in what I'm seeing.
I'm just curious.

Thanks in advance,
Karen
www.karenlatorre.com

>you may visit our joint report, Smart.Conseil and I,
>on the use of different "Thermal Expansion Coefficient"
>using different methods, including Winkelmann's ;
>
>http://www.flickr.com/photos/potier/46606941/
>
>There are more interesting results concerning
>glaze behaviour at :
>
>http://www.flickr.com/photos/potier/page3/

_________________________________________________________________
New Windows Live Hotmail is here. Upgrade for free and get a better look.
www.newhotmail.ca?icid=WLHMENCA150

Dan Semler on sat 21 jul 07


Hi Karen,

It is true that one of the problems of glaze theory and =20
specifically the application of glass theory to glazes is the degree =20
to which glazes a completely melted. One thing we often assume in =20
calculation is that COE is a constant for an oxide. Its not really so =20
though within certain temperature ranges and glaze compositions its =20
close enough for a first approximation. Actual measured and calculated =20
values for COE of a glaze a very unlikely to match. However, for well =20
melted glazes calculated COE estimates are useful in giving an =20
indication of likely behaviour. Deltas between values for gloss glazes =20
are the best examples of this. The temp range over which the values =20
are derived is important as it indicates the range over which they =20
heated the samples, or at least the range over which they considered =20
the data. If there are substantial changes in linearity outside this =20
range the numbers won't reflect it. This is something of a finer point =20
as in practical terms few of the materials appear to exhibit =20
substantial non-linearities over the ranges or temp. There are however =20
dependencies on composition of the glazes for some materials. Boron is =20
one example.

There is some description of the methods used for determining the =20
COEs in the various papers published by the orginal researchers. Get =20
your library to get you copies of Winkelmann & Schott, Appen, English =20
and Turner's papers. You'll see what kind of glazes they use and how =20
they go about measuring the values. They give absolutely all the =20
detail but a bit.

All that said I still need to think more on what might be at play =20
in your tests. I will certainly be interested to see what happens with =20
the tests on clay.

Thanx
D