David Hewitt on tue 23 mar 99
Adding Colouring Oxides can reduce Crazing
At the end of February Mike Bailey and I posted a piece which detailed
the results of some tests which we had been doing. This involved
progression blends of Red Iron Oxide, Copper Oxide and Cobalt Oxide into
a glaze and applying this to three different clay bodies, a buff
stoneware, a white stoneware and a porcelain. The base glaze tended to
craze, particularly on the porcelain body. As the % of colouring oxide
increased it was clear that the crazing decreased.
Firstly, we were interested to know if this was what other potters had
experienced.
We were grateful for a number of off-list replies from:-
Ellen Baker
Tom Buck
Khairraj Seepersad
Autumn Downey
Peter Atwood
There was general agreement that this was the case.
Secondly, we were interested in noting that this reduction in crazing
conflicted with the increase in calculated expansion with the addition
of the colouring oxides. This calculation of course depended on the
figures used for these colouring oxides. English & Turner did not
produce coefficients for Fe2O3, CuO and CoO so we had been using figures
gleaned from elsewhere and incorporated in what we called a 'Hybrid
English & Turner' set. We were all too aware of the problem of mixing
such sets of figures. If our experience is, as described above, exactly
what others have experienced, then this would support the conclusion
that the coefficients that we have been using are wrong.
Expansion coefficients for colouring oxides have been produced by Mayer
& Havas, but they did not also include figures for the oxides in the
base glaze ingredients, such as K2O CaO etc. The only ceramist that we
know of, who has produced coefficients for essentially all ingredients,
is Appen. Using his figures we do get a decrease in glaze expansion with
the addition of these oxides.
This might suggest that Appen might be the best set of coefficients to
use for any glaze Expansion calculation. We note, however, that
resultant Appen glaze coefficients all tend to be higher than English &
Turner. In our earlier work on expansion, see 'Calculating Crazing' on
either of the following sites:-
http://www.dhpot.demon.co.uk
http://digitalfire.com/education/people/hewitt.htm
we preferred English & Turner from all the others listed because from
our limited tests we found that the glaze coefficients seemed to relate
better to the clay body coefficients that we had available to us.
Also, many users of glaze calculation programs such as Insight have now
become familiar with a level of glaze coefficient which might or might
not craze. To make a step change in his level would no doubt cause some
confusion. Another approach would be to introduce Appen's figures into
English and Turner. This gives expansion figures that are much closer to
observed results.
On looking back at some earlier tests using MnO2 and Cr2O3 as colouring
oxides, it is clear that these also have reduced crazing. Appen does
give a coefficient for MnO as 105, which, from our tests would appear to
be high. We are, therefore, suggesting a lower figure. Appen does not
give a figure for Cr2O3, but we have included what we think would be
appropriate, based on our observations.
Appen E & T The figures I
Proposed have been using
% Mol %wt x 10-8 %wt x10-8
x10-7
Fe2O3 50 5 10.4
CuO 30 3 7.3
CoO 50 5 14.7
NiO 50 5 13.3
FeO 55 5.5 13.3
Cr2O3 - 5.7 17
MnO2 105 6 5.7
Any comments on these would be welcome.
For completeness the recipe of the base glaze in question is repeated
below;
The base glaze recipe was:-
Soda Feldspar 450
China clay 50
Bentonite 20
Lithium Carbonate 20
Whiting 370
Zinc Oxide 50
Flint 40
-----
1200
Fired to cone 6 oxidation
Analysis:-
K2O .03 Al2O3 .19 SiO2 1.78
Na2O .10 Fe2O3 .00 TiO2 .00
CaO .70
MgO .01
Li2O .05
ZnO .11
Coeff. of Expansion 6.34 x10-6/oC Linear English & Turner
Calculation using Appen's figures are available in our GLAZE WORKBOOK
glaze calculation file ( see web sites listed above) and in MATRIX ( see
web site
www.tekotago.oc.nz/art/matrixgoweb/default.htm)
--
David Hewitt
David Hewitt Pottery ,
7 Fairfield Road, Caerleon, Newport,
South Wales, NP6 1DQ, UK. Tel:- +44 (0) 1633 420647
FAX:- +44 (0) 870 1617274
Own Web site http://www.dhpot.demon.co.uk
IMC Web site http://digitalfire.com/education/people/hewitt.htm
Michael Banks on wed 24 mar 99
Hello David,
I appreciate your efforts on raising awareness on this effect. I personally
don't think that the oxide coefficients of thermal expansion (CTE) are the
problem. I believe anti-crazing effects are due to two uncontrolled
parameters other than their combined CTE's. The parameters are:
1. formation of low CTE minerals in the glaze lattice &
2. improvements in tensile strength due to the introduction of high
coordination number transition elements into the glaze lattice
I got a clue about this a few years ago when mixing quantities of red, pink,
violet and crimson stains with low temperature frit-based clear gloss
glazes. The base glaze crazed on the test body when the colourants were
absent. Crazing reduced noticably or disappeared when Cr-Sn, Mn-Al and Cd-Zr
colours were introduced. These commercial colours are synthetic spinel
minerals or other stable compounds which have very low CTE which are
essentially unrelated to the CTE of the constituent oxides. The very low
CTE's of spinels are due to the nature of the internal bonds and cubic close
packing of the constituent atoms. This extraordinary stable structure also
gives rise to their other desirable features such as resistance to
dissolution and refractoryness.
Spinels and other stable minerals based around zirconia and tin, among may
other elements, form naturally whenever transition elements are introduced
into glazes. This is why tin oxide or titanium dioxide for example, can
have significant effects on colour. Both of these are tetravalent ions
which do not form silicates easily. They form spinels more readily, which
have a greater effect on the CTE of a glaze than would be expected from the
constituent oxides. Cobalt probably also falls into this class as it is
known to form a stable Co-Al compound.
Tensile strength data for glazes is hard to come by, but metallurgical
experience shows that defects in metal crystal lattices can be cured by
additions of tiny amounts of other elements which result in great increases
in strength. Glazes appear to be analogous to this and small additions of
some oxides definitely improve strength. This effect is apparently due to
filling of octahedral and tetrahedral sites, holes or defects in the glaze
lattice by atoms or oxide molecules, so that glaze elasticity and resistanct
to tearing apart = tensile strength is enhanced. It kind of works like
grog.
Obviously if you have two glazes with identical CTE, but different tensile
strengths, then the stronger glaze will resist crazing to a greater degree.
My experience is that this is a non-trivial effect and that certain
colouring oxides increase glaze strength to a large degree. Tin, chromium
and zinc seem to be examples and I suspect many others such as cobalt,
nickel and iron for example.
Unfortunately tensile strength numbers and the presence of low CTE mineral
inclusions are not amenible to simple glaze calc programs. Stiff bikkies
potters!
Michael Banks,
Nelson,
New Zealand
"The only place where success comes before work is a dictionary".
- Vidal Sassoon
----------------------------Original message----------------------------
Adding Colouring Oxides can reduce Crazing
(Snip)......If our experience is, as described above, exactly
what others have experienced, then this would support the conclusion
that the coefficients that we have been using are wrong.
Autumn Downey on thu 25 mar 99
Hello David,
Was interested in your last posting re-crazing and also read the article
that you and Mike Bailey had written. (Which is not to say that I absorbed
it all, but did get the drift.)
I'm wondering what it would take to interest a
ceramics/scientific/engineering? company in
doing a new batch of expansion tests. I guess unless there's a profit
motive, it's not likely to occur. All the same, it would seem that many of
the figures you are using were developed a long time ago. Would recent
advances in physics, chemistry, measurement, etc make it any easier to do
these tests again? I can't imagine how they were done to begin with!
Autumn Downey
downeya@internorth.com
Yellowknife, NWT, Canada
Michael McDowell on thu 25 mar 99
Michael Banks made some interesting suggestions about why this anti-craze
effect of additions of coloring oxides may be taking place, even though the
thermal expansions of the oxides alone may be high, leading our glaze
calculations to predict greater crazing. Michael suggests two effects in
particular:
"1. formation of low CTE [coefficient of thermal expansion] minerals in the
glaze lattice &
2. improvements in tensile strength due to the introduction of high
coordination number transition elements into the glaze lattice"
I did my best to follow his discussion, and basically it seemed quite clear,
if technical. I think I agree with him, though I'm not enough of a scientist
to know for sure that I do. There may be other specific interactions at work,
but I'm sure we agree that it is more likely the "assumption of linearity" in
calculations of thermal expansion that is in error, not so much the
calculations of individual oxides CTE's.
This "assumption of linearity" is that we can simply predict the thermal
expansion of a mixture of separate oxides by adding together the expansion
coefficients of each one times the proportion of each one in the mix. This
assumes that the oxides only "mingle" together in the glaze. When they happen
to "interact" to create more complex molecules, "non-linearities" can result.
Mr. Banks suggests a couple of likely sources of such non-linearities. I'm
sure there may be others. I'm not so much interested in the precise mechanisms
as I am in the recognition that such effects do occur.
Actually, I don't think that there are many who advocate the use of
calculations to approximately predict thermal expansion of glazes who are not
aware that "linearity" is a flawed assumption. That's why you will always hear
these advocates suggesting calculation as "only a guide to experimentation".
However flawed, linearity is still a useful assumption. Our understanding of
the interactions that can take place between oxides in a glaze mixture is
limited. Based on my limited observations I would guess that the non-linear
effects of oxide interactions occur with relatively small amounts. As the
proportional additions of colorant oxides increase, the greater part of those
additions only mingle in the glaze melt. So where percentage additions are
larger, the calculations may be more accurate at predicting the effect on
crazing. I am only guessing here though. I would be interested in hearing more
from David Hewitt about the range colorant additions over which he observed
these anti-crazing effects.
Michael McDowell
Whatcom County, WA USA
mmpots@memes.com
http://www2.memes.com/mmpots
David Hewitt on sun 28 mar 99
Michael,
Thank you for your pertinent comments. While I find expansion
calculations for glazes most useful, particularly in indicating in which
direction a recipe change may be expected to affect the expansion, I do
not take the figures as gospel. Without the facility of a dilatometer to
actually measure an individual glaze, however, a calculated figure is
certainly a valuable tool.
As you say, research work has shown that the coefficients of different
oxides in a mix may vary to some degree according to the proportions of
the oxide in question. As I understand it, instead of a straight line
one gets a curve and the figures given, say by English & Turner etc. are
an average. You use the word 'flawed', but I would suggest that the
degree of error is quite acceptable for the use to which we put the
figures.
Some ceramists have taken this into account by varying the coefficient
according to the amount of certain oxides in a glaze. For example, F. P.
Hall refers you to a graph to determine the coefficient for SiO2. Also
Appen has variable formulae for determining the coefficients for SiO2,
B2O3, TiO2 and PbO. These formulae take into account other oxides in the
glaze. Details can be seen through my web site 'Pottery Techniques /
Calculating Crazing' or directly on the IMC Insight web site where this
report is based.
All this, therefore, to my mind, adds weight to the use of figures based
on Appen's data for colouring oxides, as the resultant glaze calculation
figures relate better to our observed results. Accepting all the
limitations to which you refer, I feel sure that you would also agree
that expansion calculations are a useful tool and hence, if one is to
use this tool, then the best available figures should be used. Hence our
proposition that they should be based on Appen's figures.
You ask about the % of colouring oxides added. The amounts are as
follows and in all cases the higher the amount the less the crazing.
Cobalt Oxide 1/4%, 1/2%, 3/4%, 1%
Copper Oxide 2%, 4%, 6%
Red Iron Oxide 4%, 8%, 12%, 16%
Chrome Oxide 1%, 2%
Manganese Dioxide 4%, 8%
David
In message , Michael McDowell writes
>----------------------------Original message----------------------------
>This "assumption of linearity" is that we can simply predict the thermal
>expansion of a mixture of separate oxides by adding together the expansion
>coefficients of each one times the proportion of each one in the mix. This
>assumes that the oxides only "mingle" together in the glaze. When they happen
>to "interact" to create more complex molecules, "non-linearities" can result.
>Mr. Banks suggests a couple of likely sources of such non-linearities. I'm
>sure there may be others. I'm not so much interested in the precise mechanisms
>as I am in the recognition that such effects do occur.
>
>Actually, I don't think that there are many who advocate the use of
>calculations to approximately predict thermal expansion of glazes who are not
>aware that "linearity" is a flawed assumption. That's why you will always hear
>these advocates suggesting calculation as "only a guide to experimentation".
>
>However flawed, linearity is still a useful assumption. Our understanding of
>the interactions that can take place between oxides in a glaze mixture is
>limited. Based on my limited observations I would guess that the non-linear
>effects of oxide interactions occur with relatively small amounts. As the
>proportional additions of colorant oxides increase, the greater part of those
>additions only mingle in the glaze melt. So where percentage additions are
>larger, the calculations may be more accurate at predicting the effect on
>crazing. I am only guessing here though. I would be interested in hearing more
>from David Hewitt about the range colorant additions over which he observed
>these anti-crazing effects.
>
>Michael McDowell
>Whatcom County, WA USA
>mmpots@memes.com
>http://www2.memes.com/mmpots
>
--
David Hewitt
David Hewitt Pottery ,
7 Fairfield Road, Caerleon, Newport,
South Wales, NP6 1DQ, UK. Tel:- +44 (0) 1633 420647
FAX:- +44 (0) 870 1617274
Own Web site http://www.dhpot.demon.co.uk
IMC Web site http://digitalfire.com/education/people/hewitt.htm
| |
|
