Glass defects originating from glass melt/fused cast AZS

refractory interaction
Citation for published version (APA):
van Dijk, F. A. G. (1994). Glass defects originating from glass melt/fused cast AZS refractory interaction.
Technische Universiteit Eindhoven. https://doi.org/10.6100/IR417346

DOI:
10.6100/IR417346

Document status and date:

Published: 01/01/1994

Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be
important differences between the submitted version and the official published version of record. People
interested in the research are advised to contact the author for the final version of the publication, or visit the
DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page
numbers.
Link to publication

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please
follow below link for the End User Agreement:
www.tue.nl/taverne

Take down policy

If you believe that this document breaches copyright please contact us at:
[email protected]
providing details and we will investigate your claim.

Download date: 14. Aug. 2020

Glass Defects Originating from Glass
MelUFused Cast AZS Refractory
Interaction

PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de

Technische Universiteit Eindhoven. op gezag van
de Rector Magnificus, prof. dr. J.H. van Lint.
voor een commissie aangewezen door het College
van Dekanen in het openbaar te verdedigen op
woensdag 8 juni 1994 om 16.00 uur

door
Franciscus Arnoldus Gerardus van Dijk
geboren te Leende

druk: wibro disscrtatiedrukkcrij, helmond.

oit proefschrift is goedgekeurd door de promotoren:
prof.dr.ir. H. de Waal
en
prof.dr. F.J.J. van Loo
Table of contents.

Summary. 1
1. Introduction. 4
1.1. Problem definition. 4
1.2. Objective. 4
1.3. structure of the study. 4
1.4. Industrial glass melting process. 5
1.5. Description of problem. 9

2. Glass defect generating mechanisms at glass/fused

cast AZS. 14
2.1. Literature review. 14
2.1.1. Mechanisms of bubble generation from
refractory linings. 14
2.1.2. Mechanisms of knot generation. 16
2.1.3. Discussion. 18
2.2. Validation of an electrochemical mechanism. 20
2.2.1 Introduction. 20
2.2.2. Description of the electrochemical
mechanism. 21
2.2.3. Electrochemical measurements. 24
2.2.3.1. Rate-determining step of the
internal reaction. 30
2.2.3.2. Shift of the location of oxygen
formation. 36
2.2.4. Discussion and conclusions. 46

3. cation diffusion into fused cast AZS. 55

3.1. Introduction. 55
3.2. Experimental procedures. 56
3.2.1. Profiles of concentration of the cations
in the glass melt and the glassy phase
of the AZS. 58
3.2.1.1. Measurement method. 58
3.2.1.2. Results of measured concentration
 profiles. 60
3.2.1.3. Summary of the measured
concentration profiles in AZS. 70
3.2.1.4. Discussion of measured
concentration profiles in the AZS. 73
3.2.2. Change in glass melt composition by
diffusion into AZS. 77
3.2.2.1. Introduction. 77
3.2.2.2. Measurement method. 78
3.2.2.3. Results of glass analysis after
refractory interaction and
determination of e.e.h. values. 82
3.2.2.4. Discussion. 96

4. Effect of glass composition and temperature on glass

defect potential originating from refractory. 104
4.1. Introduction. 104
4.2. Results of measured cation diffusion. 105
4.3. Practical application of the cation diffusion
rules by a glass furnace model. 111
4.4. Discussion. 116

5. Bubble formation during the laboratory-scale tests. 123

5.1. Introduction. 123
5.2. Experimental method. 123
5.3. Test results. 124
5.4. Discussion. 130

6. Knot formation mechanism and characterization. 135

6.1. Introduction. 135
6.2. Knots in TV-screen glass products. 136
6.3. Dissolution of knots. 141
6.4. Concentration profiles of AZS-to-glass
interface. 148
6.5. Chemical composition of the interface glass
melt/fused cast AZS. 152
 6.6. Examples of knots in industrial TV-screen
glass production and their sources. 157

Annex l . 161
Annex 2. 162
Annex 3. 171
Annex 4. 174
Annex 5. 174
Annex 6. 177
 1

SUMMARY

The interaction of glass melts and refractories in industrial

glass melting tanks has a large impact on the resulting glass
quality. The mostly applied type of refractory in the glass
industry, in direct contact with the glass melt, is fused cast
AZS (Alumina Zirconia silicate). Due to the interaction of the
glass melt with fused cast AZS refractory, different kinds of
inhomogeneities like stones, knots and bubbles can be generated.
The objective of this study is to find, for fused cast AZS in
contact with a glass melt, the mechanism of knot and bubble
formation and the parameters that influence this mechanism.
The ultimate objective is a (semi) quantitative model for glass
defect prediction from fused cast AZS in contact with TV screen
glass melts to be able to choose conditions of lowest glass
defect formation rates.
Most of the bubbles arising at temperature levels above 1400 0 C
are initially oxygen bubbles, originating from an electrochemical
mechanism, acting in the refractory in contact with molten glass.
When molten glass is brought into contact with fused cast AZS,
the diffusion of cations from the glass melt into the AZS
transports a positive charge into the AZS interior. The diffusion
of the cations is caused by the lower partial Gibbs energy of the
glass melt cations in the high AI 20) containing glassy phase of
the AZS refractory. The positive charge due to the cation
diffusion is balanced by electrons moving from the glass melt/AZS
interface into the AZS interior, resulting in an oxidation of the
oxide ions of the melt at the AZS/glass melt interface. The
electrons reduce the polyvalent ionic impurities (mainly iron) of
the fused cast AZS. The rate of oxygen gas formation at the
refractory/glass melt interface, as a result of the redox
reaction, is determined by the diffusion rate of cations of the
glass melt into the glassy phase (binding phase) of the fused-
cast AZS.
The bubbles from this source arise, as mentioned, in the
interface layer between fused-cast AZS and the glass melt. The
 2

interface consists of a vitreous layer of products resulting from

reactions between fused-cast AZS and glass melt, with Zr02-
nodules at the refractory end. A bubble generated in this
interface area forces its way through this layer into the glass
melt, sometimes squeezing part of this Al 20 3 rich layer into the
bulk of the melt. The high Al 20 3 -content of the vitreous section
of the interface gives these "particles" such a high surface
tension that they become spherical knots in the glass melt.
All this implies that one single mechanism accounts for the
formation of both bubbles and knots at the interface glass
melt/fused cast AZS.
The rate determining element of this process, the cation
diffusion, has been measured for the oxides essential for a
present-day TV-screen glass composition (Li 20, Na 20, K20, BaO,
SrO, CaD and MgO). The effect of Li 20 on the oxygen bubble
formation is very high, including not only its own contribution
but also causing other oxides to have a larger effect than the
effect of the same oxides for a glass without Li 20. By far the
highest effect of the other oxides, apart from Li 20, is that of
K20, particularly in the range around 1350 0 C (bottom temperature
of the melting tank). The formation rate of oxygen bubbles per
mole per cent of the oxides under examination is in decreasing
tendency: Li 20, K20, Na 20, BaO/SrO, CaD and MgO.
The diffusion rates of the cations of these oxides in the AZS
and, consequently, the bubble formation potentials effected by
the composition of the glass and the temperature, have been
computed using a model for a simplified version of an industrial
melting furnace with assumed realistic glass melt/fused cast AZS
contact temperatures. The differences in the glass defect
potential of the interaction between fused-cast AZS and different
glass melts computed by this method, agrees with the amounts of
bubbles and knots found in industrial produced TV glasses.
Laboratory-scale tests show a rise in number, in oxygen content
and in diameter of the generated bubbles, at increasing
temperatures (from 1350 to 1500 0 C).
The chemical composition of the core of the knots, originating
 3

from the interface between glass melt and fused-cast AZS remains
stable during dissolution in the bulk glass melt until just
before complete dissolution. This chemical composition provides
us with a method to derive the temperature of their origin, since
the chemical composition of the interface depends mainly on the
chemical composition and the temperature of the glass melt.
 4

1. Introduction.

1.1. Problem definition.

The interaction of glass melt and refractories in the industrial

glass melting tanks has a large impact on the glass quality. Due
to the interaction of the glass melt with refractory materials,
different kinds of inhomogeneities like stones, knots, cords and
bubbles are generated. It is estimated that, depending on the
specific product and required glass quality, up to 10 % of the
industrial glass production in continuously operating furnaces is
rejected due to these kinds of defect sources (Lit. 1).

1.2. Objective.

The objective of this study is to find for the most applied type
of refractory, fused cast AZS, the mechanism of knot and bubble
formation and the parameters that influence this mechanism. The
ultimate object is a (semi) quantitative model for glass reject
prediction from fused cast AZS (alumina-zirconia-silica) in
contact with TV screen glass melts, to be able to choose
conditions of lowest defect rates.

1.3. The structure of the study.

A short description of the glass melting process and glass reject

problem is given in this chapter.
A literature review on the generation of glass defects is given
in chapter 2. This chapter also describes the measurements and
results to validate the mechanism for bubble and knot formation
during the interaction of fused cast AZS with a glass melt. The
diffusion of cations, into the fused cast AZS, appears to be the
rate determining step in the formation of bubbles and knots.
Often, in this study, fused cast AZS will be simply referred to
as AZS.
In chapter 3, the experiments and results on the diffusion of the
 5

most important cations of a TV screen glass will be presented.

Although the cations are diffusing into the AZS, the name of the
oxide of the cation is used to express the amount or rate of
diffusion. The name of the oxide is used because the analytical
measurement of the glass composition is also in oxides. with the
results of the experimental work described in chapter 3, a simple
model has been prepared which is presented in chapter 4. With
this model it is possible to estimate the effect of the glass
composition on the glass reject due to the interaction of the
glass melt with fused cast AZS.
In chapter 5 the behaviour and nature of bubbles in the melt,
arising in laboratory scale tests, due to interaction of glass
melt and fused cast AZS, are described.
In chapter 6 the mechanism for knot formation, their chemical
composition and dissolution kinetics are presented. A completely,
new method to find the source of knots is also presented in this
chapter.
with this new method the temperature of the origin of the knots
in the furnace can be determined after the chemical composition
of the knot has been analyzed.

1.4. Industrial glass melting processes.

The industrial production of glass from raw material batch is

generally a continuous process with a high degree of
automatization.
There are many different kinds of glass products. The main glass
products are: - container glass
flat glass
domistic glass (tableware)
insulation glass fiber products
crystal glass
textile and reinforcement fibers.
- glass for electronic applications.
Further there is a wide range of special glass compositions like:
- borosilicate glass
 6

- opal glass
- potassium glass
- glass ceramics.
All these glasses have a different chemical composition with Sio2
as main glass forming component.
The glass output of continuous glass melting furnaces varies from
5 tons/day, up to 800 tons/day. The melting area of such furnaces
varies from a few m2 , up to 300 m2 •
In most cases, fossil fuels are the energy suppliers. In some
cases electric boosting or all electric heating is applied. When
fossil fuels are used, the hot flue gases are used to preheat the
combustion air (except when pure oxygen is used instead of air).
For small production units «100 tons/day), in most cases a
metallic recuperator is used, for larger units a more energy
efficient regenerative system is generally applied.
Often the furnace consists of a melting end, a working end and
one or more forehearths and feeders. For the glass melting tank,
nowadays, fused cast refractory materials are used. The melting
end is kept on a temperature of 1400 till 1620 °C, depending on
the glass composition, throughput and required glass quality. A
well mixed batch of raw materials combined with cullet (from
rejected glass or recycling waste glass) is continuously charged
to the melting end. The batch floats on the glass melt and is
heated by the radiation of the furnace crown and the flames from
above and the transfer of heat from the hot glass melt from
underneath.
During the heating process of the batch, a lot of chemical
reactions take place: starting with solid state reactions between
particles of the raw materials, which form low melting eutectic
phases. The batch particles like sand or feldspar dissolve in
these reactive melts, by diffusion processes. Because part of the
raw materials are added as carbonates and often water is used in
the batch making process, the melting of the batch is accompanied
by dissociation reactions which lead to the formation of gases
like carbon dioxide and water vapour. The largest part of these
gases and the air entrapped between the batch particles are
 7

released from the batch before it is completely melted. Although

a small part will be trapped in the glass melt. A fining agent is
used (mostly sulfate, antimone (V) oxide or arsenic (V) oxide) to
remove these gas inclusions (bubbles) by blowing them up at a
certain temperature level, thereby increasing their rising
(buoyancy driven) velocity in the moderately viscous melt.
The complete dissolution of all solid particles and the
homogenisation of the glass are other main processes which have
to take place in the melting end. The quality of the glass
product is strongly dependent on the conditions during this
melting process. For the quality of the glass, the main
parameters are: process stability, melting end temperature,
temperature distribution, mean residence time, residence time
distribution, batch composition, redox state, quality of the
refractory materials and the aging of the complete furnace
system. In most furnaces, the glass leaves the melting end
through the throat, after a mean residence time of 25 to 60 hours
for most of the industrial processes, into the so called working
end. In the working end the glass is cooled down and conditioned
(to obtain a melt with an uniform temperature) before the melt
flows to the forehearth(s). There the melt will be cooled down
and conditioned to the temperature related to the viscosity
needed for the forming process of the required shape and weight.
In the forehearth, often stirrers are located to increase the
chemical and the temperature homogeneity. In the feeder the glass
melt output is controlled with the aid of a plunger (sometimes in
combination with a tube) and the glass melt output flows through
a hole at the end of this feeder. This glass is cut in a constant
interval of time. The portions of glass made in this way are
called gobs. These gobs are made with a narrow weight and size
range, necessary for exact article shape and dimensions.

The layout of a typical, so called regenerative cross fired,

industrial furnace is illustrated by figure 1.1. The glass
quality achieved with such a process is mostly sufficient for
articles with relatively modest quality requirements.
 8

figure 1.1 Cross fired regenerative container or T.V. glass

furnace.

regenerator

doghouse or
charging end

/
feeder
"
horizontal cross section

burner ports

Regenerator
for cyclic
air perheating
checker of and flue gas
heat storing cooling.
bricks Air and flue
gas direction
change in
cyclic process.

vertical cross section

 9

For more general information about glass, glass melting, furnaces

and glass defects the standard literature of Scholze (Lit. 2),
Trier (Lit. 3), Noelle (Lit. 4) and Jebsen-Marwedel and Brueckner
(Lit. 5) is recommended.

1.5. Description of the problem.

In many areas of glass manufacturing production losses due to

glass defects such as bubbles, stones and knots are very high.
- Bubbles or blisters are gaseous inclusions in the glass.
- stones are solid unmolten or recrystallized material inclusions
in the glass.
- Knots are vitreous particles of a deviating chemical
composition compared to the bulk and consequently, a different
index of refraction, which brings about a local lens effect.
The vitreous particle may contain inclosed crystals like
recrystallized zirconia, nepheline or leucite.
An example, for high production losses due to these glass
defects, is the colour television screen industry, where
manufacturers are rejecting screens because of the presence of
extremely small bubbles, stones or knots. For the production of
screens for computer monitors, spherical bubbles with a
volumetric diameter of 0.35 mm are rejected. For the production
of normal consumer TV screens products with spherical bubbles
larger than 0.40 mm in small screens or 0.6 mm for some areas of
the larger TV screens will be rejected.
All visible knots and stones will lead to rejection.
These rigid customer requirements, for the TV screen production
leads to losses due to such glass defects of 5 to 25 %, depending
on the size of the screens. For the production in one glass
furnace with a capacity of approximately 180 tons/day, 1%
rejection means a loss of us $500,000 on an annual basis. Most
manufacturers have several of these kinds of melting tanks for
the production of TV screens. This means that the number of glass
defects has a large influence on the profitability for all TV
 10

screen manufacturers but also for other manufacturers of high

quality glass like float glass for automotive windshields and
mirrors, reinforcement fibers etc.
There are a large number of possible sources of glass defects.
For stones the following are the most common (Lit. 6):
- pollution of raw materials.
- pollution of cullet.
- refractory - clay-type refractory.
sintered AZS and zircon or cements.
- fused cast AZS.
For knots, the fused cast AZS is the most common source.
For bubbles the following sources are common (lit. 7):
- the primary melting of raw materials.
- incomplete fining process.
- reactions of the melt with refractories and metals.
- reboil (The nucleation and growth of bubbles in previously
bubble free glass, by overheating or sudden changes
in redox state).
lapping-in bubbles, by moving parts (e.g. stirrers,
plunger, tube or gobber) in the feeder.
- bubbles, originating from organic particles.
For the exact determination of the origin of glass defects 1) the
appearance, 2) generally the shape, 3) the location in the
article, 4) the mineralogical and 5) chemical analysis of the
defect has to be available. Further, it is essential to be well
acquainted with the refractories installed, furnace conditions,
fluctuations in furnace conditions, furnace settings, stirrer
settings and gob making. In such cases, often the origines) of
the glass defects can be found.
During production of glass for products which require a high
glass quality, the reject originating from refractory is
considerable: causing approximately 50% of all the glass defects.
In the start up period of a furnace, with extremely high glass
rejects (Lit. 8), practically all the products which are rejected
show defects originating from refractory/melt interactions.
Therefore this study will deal with the glass defects generated
 11

at the glass melt/fused cast AZS interface. Fused cast AZS (see
table 1.1) is chosen because it is by far the most applied fused
cast material for the melters in the glass industry today.

Table 1.1

Fused cast AZS.

fused cast fused cast

AZS 32/33 AZS 41
Chemical composition in weight %.

32-33 40-41

50-51 45-46

12-13

1 1

0,2 0,3

Crystallographical composition in weight %.

Zirconia (Baddeleyite) 32,0 41,0

Corundum 47,0 42,0

Glass phase 21,0 17,0

The glassy phase of AZS refractory, consists of about 70 weight%

Si021 4 weight% Na 20, 22 weight% Al 20 3 , 3 weight% Zr02 (Lit. 1),
0.3 weight% Fe 20 3 and O. 3 weight% T i0 2 (Lit. 9).
The fused cast AZS is produced by melting the raw materials in an
electric arc furnace at about 2200-2400 °c. The main raw materials
 12

are about 50% recycling of fused cast AZS from direct reject
blocks or over cast and the primary raw materials: zirconium
silica, zirconia, alumina and soda ash. The melt is treated with
oxygen to oxidize zirconium oxy-carbo-nitride or other reduced
constituents, produced by the reaction with the graphite
electrodes during melting. The molten material is then cast into
molds to, or very near to, the final shape of the block desired
(lit. 8). The molds are mostly made from a special kind of sand.
The mold with the fused cast ceramic is cooled down very slowly in
order to anneal (free of thermal stresses) the fused cast product.
The produced fused cast AZS contains baddeleyite (Zr02) and
corundum (Al 20 3 ) crystals which are kept together by a glassy
phase. The material is not very homogeneous. Crystal size,
chemical composition of the glassy phase and porosity differ
locally in one block.

Figure 1.2
Typical example of the microstructure of fused cast AZS (32/33),
magnification 500 X.

Baddeleyite Zr02

Corundum Al 20 3

Glassy phase
 13

Literature references chapter 1.

[1] R.G.C. Beerkens, A.J. Faber;
Refractories in contact with molten glass;
TNO Industrial Research TPD-GL-RPT-93-091.
[2] H. Scholze;
Glas, Natur, Struktur und Eigenschaften;
Springer-Verlag 1977.
[3] W. Trier;
Glasschmelzoefen, Konstruktion und Betriebsverhalten;
Springer-Verlag 1984.
[4] G. Noelle;
Technik der Glasherstellung;
Verlag Harri Deutsch 1979.
[5] H. Jebsen-Marwedel, R. Brueckner;
Glastechnische Fabrikationsfehler;
Springer-Verlag 1980.
[6] G. Duvierre, A. Krings, E. Sertain;
Defects and their origin in glass;
Glasteknisk Tidskrift 45 (1990) p. 63-70.
[7] E.L. Swarts;
Gases in glass;
Ceram. Eng. Sci. Proc. 7 (1986) p. 390-403.
[8] A.D. Davis, L.L. Cureton;
Start-up and surface blistering of fused-cast refractories;
Ceram. Eng. Sci. Proc. 8 (1987) p. 276-284.
[9] M. Dunkl;
Studies on the glassy and reaction phases given off by fused
cast AZS blocks and their effects on the glass quality;
Glastech. Ber. 62 (1989) p. 389-395.
 14

2. Glass defect generating mechanisms at glass/fused cast AZS.

2.1. Literature review.

2.1.1. Mechanisms of bubble generation from refractory linings.

Refractory porosity.
Fused cast AZS bodies have a detectable porosity. Pristine fused
cast AZS material has no open porosity but there are some closed
voids and pores which arise from the manufacturing process. These
pores are filled with gases. The amount of pores in the material
is strongly dependent on the applied production process. The gases
in these pores have an air like composition, with nitrogen and
oxygen as main components (Lit. 1) and some carbon dioxide (Lit
2) •
When the refractory is in contact with glass, the refractory
material will react with the glass melt and will slowly dissolve
in this melt: the closed pores will be opened and then release
their gas content, forming bubbles.
Impurities in the refractory.
Impurities, which can oxidize, are for example elemental carbon
originating from graphite electrodes of the fusion casting
operation, sulfur or zirconium carbide originating from raw
materials of the AZS. The dissociation/oxidation of
nitrides/oxynitrides, produced by reactions with the graphite
electrodes during melting of the AZS, could give nitrogen gas.
At first, during the heating of the fused cast AZS at temperatures
higher than 1400 °e, these oxidation reactions, involving the
formation of nitrogen, carbon dioxide, carbon mono-oxide (Lit. 1 &
3) and/or sulfur dioxide (Lit. 4) are a cause of bubble formation.
When fused cast AZS is cooled down and heated up again, generally
02 bubbles are formed. Generally, at temperature levels higher
than 1400 °e these O2 bubbles are formed every time when the
temperature is increased. (Lit. 1 & 3). As the material is
cyclically cooled down and heated up again, at every heating up
period, new oxygen bubbles are generated from the AZS in glass
 15

melt contact but also in the absence of glass melt contact. In the
cooling down period, oxygen is resorbed again, this can be caused
by the redox reactions of polyvalent ions (as impurities) in the
refractory material. The oxygen is resorbed by the polyvalent ions
because generally the redox reaction of a polyvalent ion shifts to
the oxidized side when the temperature is decreased. The fused
cast AZS contains as impurities «0,3 weight%) the oxides of
polyvalent ions like iron (Fe2 + /FeJ+) and titanium (Ti 2 + , Ti4+) •
For example: 4 Fe2 + + O2 -+ 4 Fe 3 + + 2 0 2-
Electrochemical reactions.
When a refractory (conductive for electrons, oxide ions and
cations) is brought into contact with a glass melt, an electrical
potential (emf) is generated between the refractory and the glass
melt because generally no thermodynamic equilibrium exists between
these materials.
The Gibbs energy changes, for redox reactions occurring in an
electrochemical cell may be expressed in the form of an
electromotive force. The electromotive force of the cell, or
difference in electrical potential between two electrodes, depends
on the equilibrium constant of the chemical redox reactions that
may take place in the cell, and the activities of reactants and
products. Then the Gibbs energy change is equal to the electrical
work at constant temperature and pressure. The electrical work is
equal to the product of the voltage and the quantity of
electricity (expressed in coulombs). The quantity of electrical
charge corresponding to the molar quantities indicated in the
balance chemical equation is nF, where n is the number of
electrons transferred per molecule and F is the constant of
Faraday. If this quantity of electrical charge is transported
through a potential difference of E volts, the amount of work
required is given by nFE. Since this electrical change does not
involve pressure-volume work and is carried out isothermally the
change in Gibbs energy, for a reversible process, is given by
(Lit. 16).

~G - nFE
 16

A relation between the electromotive force (emf) generated between

glass melt/refractory interior and the formation of oxygen bubbles
was found by Bossard and Begley (Lit. 5). They also showed a
positive correlation between the electromotive force and the
sodium oxide concentration difference between the glass melt and
the refractory.
A similar influence of cations in the melt on the electromotive
force, between the refractory and the glass melt, was found by
Leger, Boffe and Plumat (Lit. 6).
For zircon, (zirconia silicate ZS) Baucke and Roeth reported (Lit.
7) oxygen bubble generation at the interface refractory/glass and
they proposed a mechanism in which the 0 2- from the glass is
oxidized to 02 and iron(III) and titanium(IV) in the refractory is
reduced. The electrical charge is transported simultaneously by e-
and alkali+ species. According Baucke and Roeth, the reaction rate
or 02 formation at the interface refractory/glass melt in this
mechanism is controlled mainly by the diffusion of alkali ions
into the refractory material. The diffusion of cations is probably
caused by the difference in partial Gibbs energy between the glass
melt and the refractory of the cations. This oxygen bubble forming
mechanism will be explained more extensively in this chapter.

2.1.2 Mechanisms of knot generation.

The generation of a (aluminia rich) knot is due to one or more

forces, which drive the fused cast AZS glass phase towards the
refractory surface (above glass melt surface) and into the bulk of
the glass melt.

The following forces may be distinguished:

Gravity.
For relatively low viscosities of the glassy phase of the AZS
nearest to the refractory hot-face, glassy phase could vacate the
refractory and be replaced by the glass melt via capillary action
(Lit 6).
 17

Zirconia transformation and hysteresis.

Displacement of the fused cast AZS glassy phase could be caused by
the thermal expansion of zirconia. Between 900 and 1200 °c a
irregularity in the thermal expansion appears by the reversible
phase transition from monoclinic to tetragonal Zr02 (Lit. 8). This
alteration of crystal form, reverses the step-sided expansion
curve and may cause shrinkage (see figure 2.1). By this expansion
and shrinkage, a pumping action could occur, expelling glassy
phase from the AZS material (Lit. 4).

figure 2.1 Thermal expansion of fused cast AZS 32/33.

r.1(%]
1.4

1.2

F-AlS

0.8

0.6

0.4

"1
0+-1
500 '000 lS00
T"CI

Gas formation.
The gas forming mechanisms are described in section 2.1.1.
In the literature (e.g. 1, 3, 9 and 10), the gas formation in the
refractory and at the interface refractory/glass is found to be
the most important cause of pushing out liquid or glassy phases
(exudation) from the fused cast AZS interior.
 18

The concentration profiles of cations in the interface region of

the glass melt/fused cast AZS (see chapter 6 of this study) show
an increasing amount of Al 20 3 in the interface, going from glass
melt to fused cast AZS. A rise of AI 20) content in glass melts,
increases the viscosity of the glass melts dramatically. In
chapter 3 and 6 it is also shown that shear stress (owing to
forced convection) has hardly an effect on the concentration
profile of the interface glass melt/fused cast AZS. Convection
will only remove, as a ream, the part of the boundary layer at the
glass melt side, with a relatively low Al 20 3 content and therefore
low viscosity. The increased removal of the boundary layer at the
glass melt side due to convection, accelerates the corrosion of
the refractory and deteriorates the homogeneity of the melt.
The knots found in practice generally have a high Al 20 3 content, as
will be shown in chapter 6, therefore a force from within the
refractory has to push out parts of the transition layer between
molten glass and refractory.
This implies that mainly the bubble forming mechanism in the
refractory interior is responsible for the presence, of bubbles
and knots in the glass, originating from fused cast AZS.

2.1.3. Discussion.

The bubble forming mechanism appears to be the key for the glass
defects (bubbles, stones and knots) originating from fused cast
AZS. This bubble formation can be caused by more than one
mechanism. Investigators (Lit. 1, 3, 11), performing experiments
using fused cast AZS materials which have been heated up more than
once to a temperature higher than 1400 °c, reported the generation
of oxygen bubbles, except during the first heating period when
bubbles of other compositions were found. The bubbles generated
during the first heating period were mainly produced by the
oxidation of impurities in the fused cast AZS.
From the mechanisms for the bubble formation process mentioned in
2.1.1., only the redox reactions of iron and titanium and the
electrochemical reactions lead to the generation of oxygen
 19

bubbles.
In the test without glass contact of Dunkl (Lit. 3, 11), the
exudation at 1550 °c of the glassy AZS phase stopped after 80
hours as the AZS has been heated up for the first time. At the
first heating period the exudation of the glassy phase is caused
by bubble formation due to the oxidation of the impurities.
According to Dunkl, no bubble formation and further changes of
redox states of polyvalent ions take place after 80 hours of
isothermal treatment. A further indication which proofs that the
the change in valence of the polyvalent ions by redox reactions
takes place in a limited time, is the short dwell time (2 hours),
at constant temperature, used in the experiments with heating
cycles. This is confirmed by Ratto (Lit. 12) using temperature
cycles from 1350 to 1550 °c and reverse. A few hours, after
reaching the maximum temperature the exudation of glassy phase
stopped. However, in experiments done by Meden and Van der Pas
(Lit. 13) in fused cast AZS crucibles filled with molten glass,
fresh oxygen bubbles were still found after a isothermal treatment
of 500 hours at 1400 and 1500 ·C. This means that the proposed
mechanism for oxygen formation (Lit. 1, 3) by the redox shift of
the polyvalent ion impurities in the fused cast AZS cannot be the
only reason for the oxygen formation, certainly not in situations
where the temperature has been constant for a long period of time.
Reactions of iron in AZS at high temperatures:

4 Fe3+ + 4 e ++ 4 Fe 2 +

The existence of electrochemical reactions, for which the reaction

rates are controlled by the diffusion of cations of the glass melt
into the refractory material, would give a much better explanation
for the long duration of the oxygen bubble formation at the glass
melt/AZS interface at constant temperature levels.

4/n ~+ (melt) ++ 4/n ~+ (diffusion into AZS)

 20

The diffusion of cations from the glass melt into the AZS
transports a positive charge in the AZS. The positive charge is
balanced by electrons moving simultaneously from the glass
melt/AZS interface into the AZS interior. The electrons shift the
Fe 3 + /Fe2+ ratio and O2 is formed by the reaction, 2 2
0 - -- 02 + 4 e-
(at the glass melt/refractory interface) as long as there is
cation diffusion.
Baucke and Roeth (Lit. 7) reported the above described mechanism
for the interaction of zirconia silicate refractory with molten
glass.

2.2. Validation of an electrochemical mechanism.

2.2.1 Introduction.

The mechanism suggested by Baucke and Roeth (Lit. 7) for the

formation of oxygen bubbles for glass melts in contact with
zirconia silicate will be verified in this study, using fused cast
AZS in direct molten glass contact.

A glass melt which is brought into contact with fused cast AZS is
not in a chemical equilibrium with the refractory material. The
partial Gibbs free energy of glass melt components like sodium,
potassium, barium and strontium are lower in the high Al 20 3
containing glassy phase of the AZS material. Therefore these
elements will diffuse from the molten glass, in the form of
cations, into the AZS material.
The electroneutrality in the AZS material, can (theoretically) be
maintained: a) by the diffusion of 0 2- into the AZS material, b) by
the diffusion of A1 3 + and/or Zr4 + from the AZS material in the
glass melt or c) by the diffusion of electrons from the glass melt
into the fused cast AZS. A requirement for the last mentioned
transport is the presence of polyvalent ions in the AZS, which can
react with the electrons.
At high temperatures, the glass melt and the fused cast AZS are
electrically conductive and AZS contains, as impurities,
 21

polyvalent ions like iron and titanium.

Therefore, the charge balancing of the cation diffusion by the
electrons is possible. The charge balancing electrons are
withdrawn from the oxygen ions in the glass melt which causes the
formation of 02 particularly at the interface glass
melt/refractory.
In this chapter the proposed hypotheses of electrochemical
reactions, as the major cause for the formation of O2 bubbles, is
validated experimentally using electrochemical methods.

2.2.2. Description of the electrochemical mechanism.

In 1988, Baucke and Roeth (Lit. 7) suggested a mechanism for the

formation of oxygen bubbles at the interface of the glass melt and
ZS (zirconia silicate). Here, it will be illustrated that a
similar mechanism, in an adapted form also applies for the case of
fused cast AZS, implying a redox reaction between an oxidic melt
and fused cast AZS. This AZS contains as impurities a minor amount
of polyvalent ions (such as iron) in their higher valence states
(Fe3 +), which is frozen in, during the formation of the AZS
blocks. When the temperature is high enough and if a second redox
system is present, the polyvalent ion can be reduced:

4 Fe3 + (AZS) + 2 0 2- (m) ... 4 Fe2 + (AZS) + O2 (g,m) (1)

in which (m) stands for melt and (g,m) for the gaseous state or
physically dissolved gas in the melt.
According to Frischat (Lit. 17) the bond strength of oxygen ions
in the glass structure is so large that they may be regarded as
stable for processes like diffusion of alkali ions.
Therefore, polyvalent ion/oxygen reaction can take place in the
refractory interior, only if the conductivity of the electrons
(generated by oxidation of the oxygen ions at the surface) in the
refractory is high enough in order to migrate to the locations of
the higher-valence ions in the AZS interior.
The negative charge migration into the AZS is kept in balance
 22

(which is a constraint to keep electro neutrality conditions) by

the migration of cations from the melt into the refractory.

2 0 2- (m) ++ 02(g,m) + 4 e (diffusing into AZS) (2)

4 Fe3+ (AZS) + 4 e- (AZS) ++ 4 Fe 2+ (AZS) (3)

4/n ~+ (m) ++ 4/n ~+ (diffusing into AZS) (4)

This means the AZS should also be capable of conducting cations M.

M = K+, Na+, Li+, for example.
Figure 2.2.1 shows the reaction scheme.

figure 2.2.1

Outline of the hypothesis on the formation of 02bubbles. (7]

melt AZS refractory

melllAZS reduced reduction unchanged
Interface layer zone AZS
(electrode) (electrolyte) (electrode)

(C._.U._)
20 2- 4e' 4e-
02.J 4Fe 2 0 4Fe 3+

(CM...UM+)
4M+ 4M+ 4M+

I I
internal cell

The oxygen ions release electrons near theAZS surface which are
consumed in the 'reduction zone' between the reduced system and
the still oxidized internal redox system. The reaction front keeps
moving deeper and deeper from the glass melt into the refractory
interior. Through the intermediary 'reduced layer' electrons and
 23

cations of identical current densities are simultaneously

transported, on account of the electro-neutrality condition. The
layer between glass melt/AZS interface and the redox reaction
front in the AZS forms an internal galvanic cell with the
interface between melt and refractory as negative electrode
(anode). For this anode the oxygen ion reservoir is the melt. The
unchanged AZS is the positive electrode (cathode). The reduced
layer is acting as a cation-conducting solid electrolyte (a 'salt-
bridge' for the cation and a 'wire' for the electron
conductivity). The overall cell reaction, in this case, is the
summation of reactions (2), (3) and (4) involves reactions and
diffusion.

4 Fe3 +(AZS) + 2 OZ-(m) + 4/n M"+(m) ++

4 Fe2+ (AZS) + Oz(g,m) + 4/n M"+ (AZS) (5)

The electromotive force (emf) of the internal cell is expressed by

equation (6):

RT (ap/\AZS)' a'''o2-(m)' allnMn+(m»

~AZSlm + In (6)
F

in which a = activity of the ions in question, fm(g,m) = oxygen

fugacity at the interface glass melt/AZS, R = gas constant, F =
Faraday's constant and E~zs/m = the standard emf for this system at
the temperature T in question. Most tests in this study have been
preformed under isothermal conditions.
The internal cell resistance is determined, according to Baucke
and Roeth in Zircon (ZrSi04 ) , by the concentration c Mn + and the
mobility UM"+ or the diffusion coefficient of the cations in the
solid electrolyte, the 'reduced layer'.
It is impossible to measure the emf of the internal cell either
directly because the melt/AZS electrode is a fictitious one, or
indirectly because the electron migration through the solid
substance represents an inherent, internal short-circuiting of the
 24

cell.
The cell voltage can be measured indirectly only if the internal
cell is integrated in an experimental cell arrangement in which
electrodes are incorperated:

Ec(t) = E(t=O) - dE(t) - (ii.R;) (t) (7)

in which -(ii.~) is symbolizing the decrease of the measured

external cell voltage by the internal current and internal
resistance of the solid electrolyte.
The -dE(t) describes the change in Gibbs energy difference between
the glass melt anode and the external AZS cathode, by the
diffusion of the cations of the glass melt in the AZS, changing
the chemical composition of the glass melt and eventually also the
chemical composition at the external AZS cathode, thus changing
the dG between AZS and the glass melt during the exposure time.
The cell reaction is irreversible owing to the diffusion of
cations and electrons from the AZS surface into the reducing zone.
With a given Gibbs energy of the cell reaction (5)

dGAZS/m - 4 FEAZS/m (8)

the rate of the redox reactions in the AZS appears to be

proportional to the increase in thickness (ds/dt) of the reduced
layer (Lit. 7). In the next part of the chapter, experimental
support will be given for the thesis that the diffusion of the
cations determines the reaction rates.

2.2.3. Electrochemical measurements.

In order to validate the proposed mechanisms, the internal cell

has been examined using the equipment given by figure 2.2.2, which
comprises an AZS (ER 1681) crucible with a platinum wire (diameter
1 mm) surrounding it as a ring. In the glass melt (about 360 g of
the so called 354-glass or 395 glass or lithium glass, see annex 1
for the chemical composition) a platinum probe, with a surface
 25

area of about 800 or 20 mm2 has been positioned.

The electrochemical cell consists of a platinum electrode, the
glass melt, the internal cell with a time-dependent reduced layer
thickness set) in the AZS, the unaffected AZS having a thickness
(d-s(t» and a second platinum electrode.

figure 2.2.2
Test set up.

Pt··

- .---

glass
':--
"Pt
./'~

AZS

The increase in the thickness of the reacted (reduced) zone, may

result in a change in the resistance of the internal cell.

(9)

in which A stands for the cell surface area.

The electrical resistance of the internal short-circuiting of the
 26

internal cell is expressed by

~(t) = s(t)/AFC~U~ (10)

in which c~ is the concentration and U~ the mobility of the free

electrons in the reduced layer.
The resistance for electron transport of the unaffected refractory
is expressed as:

R' ~(t) (d-s(t»/AFC'~U'~ (11)

in which, C'~ is the concentration and U'~ the mobility of the

free electrons in the original AZS.
In the case of short-circuiting the external cell or applying an
external voltage (see figure 2.2.3), the following argumentation
for the electrical resistances in this practical arrangement can
be made: The surface of the glass melt/AZS contact is very large
compared to the surface of the glass/platinum or the platinum/AZS
electrodes, so the current densities in the refractory will be
very low compared to those at the platinum electrode contacts.
Therefore the polarization resistance to the migration of cations,
~.M+' and of electrons R..," across the melt-to-AZS interface is
presumably negligible at high temperatures and low external
current densities.
The electrical resistance of the melt: ~, is nearly constant at a
given temperature. Probably, the small increases in the
concentration of corrosion products in the glass melt, originating
from the AZS and the small decreases of cations due to the
diffusion into the AZS, hardly effects the value of Rm.

The mechanism for oxygen bubble formation suggested by Baucke and

Roeth has been experimentally tested for zirconium silicate
refractory by these authors.
In the here presented study, electrochemical measurements are
performed, in order to test experimentally whether this mechanism
is also valid for the interaction of fused cast AZS (Electrofused
 27

Cast Alumina Zirconia Silica) and molten glasses.

The electrochemical measurements applied in this study have been:
1. Measurement of the external cell voltage, E••
These measurements are carried out to illustrate that indeed an
external cell voltage exists and that this external cell
voltage becomes gradually lower by the diffusion of cations in
the fused cast AZS. When the external cell voltage drops due to
the diffusion of cations in the fused cast AZS it is important
to know if the chemical composition of the glass melt has an
influence on this drop. When this is the case the rate of the
reaction and electron transfer (8) is determined by the
diffusion rate of the cations which also determines the
formation of oxygen bubbles in the case that the proposed
mechanism is valid.
2. Cell short-circuiting.
If the proposed electrochemical mechanism is valid, it must be
possible to shift a part of the oxidation away from the glass
melt/AZS surface. In a short-circuiting test set-up a part of
the oxidation at the glass melt/AZS should be shifted to the
platinum electrode surface in the glass melt. If this
experiment is positive, part of the required electrons
necessary for the redox reactions in the AZS interior are
transported through the external circuit. In that case the
mechanism responsible for the oxygen bubble formation is of an
electrochemical nature.
Probably, the short circuiting of the cell only shifts a
relatively small part of the oxygen bubble formation from the
AZS glass melt surface to the platinum-to-melt electrode
surface. Therefore a third measurement method has been applied.
3. Electrolysis by external applied voltage U, with the outer
surface of the AZS crucible as the negative electron donating
pole and a positive pole in the glass melt.
In this way a larger part of the oxygen bubble formation at the
interface AZS/glass melt can be forced to shift to the
platinum-to-melt electrode surface and the generated electrons
are transferred via the external circuit to the outer crucible
 28

wall. At conditions in which a sufficient part of the oxygen

bubble formation is shifted, the possibility of suppression of
the bubble formation at the molten glass/AZS interface can be
proven. This can be done by counting the number of bubbles
formed at the interface AZS/glass melt. The possibility of
suppression of the oxygen bubble formation by a counter current
proves the existence of the proposed electrochemical mechanism.
Further points of interest are the influence of the applied
external voltage on the diffusion rate of the cations into the
refractory and on the electric current.

These three tests are used to prove the validity of the proposed
mechanism for the interaction of fused cast AZS with a glass melt.
In that case the diffusion of the cations of the glass melt into
the fused cast AZS should determine the rate of oxygen (bubble)
formation. Then, quantification of the diffusion of the cations
gives a possibility to predict the tendency of oxygen formation at
the interface fused cast AZS/glass melt.
As shown in the illustration of figure 2.2.3, in this experimental
study the cell has been subjected to:
1. measurement of external cell voltage, Eo;
2. short-circuiting of AZS with a platinum electrode;
3. electrolysis by external voltage U, with the AZS crucible
outerwall as the negative pole.
In accordance with the law of electro-neutrality the summation of
the current densities in the AZS have to be always zero:

A~

L i=O (12)

in other words:

A~

Li D+
M + i~ + i'~ 0 (13)
figure 2.2.3

Electrochemical cell In detail.

r------
t:---1 U h
R~ f-I-~L--R_----'_O----,
m

I
d

s -- ~~ (d-S)
Eo - - ~
...... pt

R
m
R' •.
RPt,m

pt

R ••

glass AZS
 30

in which
. •+
1M current density by cations
i .. current density of electrons in internal cell
. , ..
1 current density of electrons in unaffected AZS

2.2.3.1. Rate-determining step of the internal reaction

The complete cell reaction is the summation of the cathodic

electrode reaction in the AZS interior, the cation migration in
the AZS and two competing (if short circuited) anodic electrode
reactions: a} oxygen formation at the AZS surface and b} (if short
circuiting the external loop) at the platinum electrode in the
glass melt.

4
4 Fe3 +(AZS} + 2(1-x}02'(m,AZS} + 2x0 2'(m,Pt} + M"+ (m) ..
n
(14)
4
4 Fe 2+(AZS} + (1-X}02(g,m,AZS) + x0 2(g,m,Pt} +
n

in which 02'(m,AZS}, 02'(m,Pt), 02(g,m,AZS} and 02(g,m,Pt) stand for

oxide and oxygen in the melt at the AZS surface (m,AZS) and at the
platinum electrode (m,Pt), and x for the fraction of oxygen
formation at the platinum electrode. The melt is not completely
homogeneous, for instance at the interface fused cast AZSj glass
melt, the chemical composition, due to corrosion products, is
somewhat different from the bulk composition. This inhomogeneity
of the melt is probably not essential for this (qualitative)
study, as long as the electric current densities are not too high.
Therefore the melt will be treated in the electrochemical
measurements as a homogeneous melt.
 31

The emf for equation (14) in a homogeneous melt with the

assumption:

(15)

in which EOAzs.m.Pt is the standard potential for any value of x.

The emf cannot be measured for all values of x, due to the short-
circuiting of the internal cell.
The driving force of reaction (14) depends on x, if the partial
pressure of oxygen at the platinum electrode differs from the
oxygen fugacity in the melt. If they are identical, the emf,
EAZS,m,Pt' is independent of x and the emf of EAZS,Pl (15) with x=l will
be identical with the value for the internal cell, EAZS,m' with x=O
(6) •
The resulting external cell voltage which can be measured, is
Ec(t) EAZSJPl(t=O) - AE(t) - (i;.~) (t) (16)
EAZSlm (t= 0) - AE (t) - (i j • ~) (t)

and its time-dependence

-d (AE + i;.~) /dt (17)

dt

provides details on the time-dependency of the effect of the

'internal iR drop' of the internal cell, and the change in the
Gibbs free energy between the glass melt and the AZS outer
crucible wall.
The migration of cations (4) may change the measured emf value
during this process, whatever the value of x will be.
 32

Measurement of the external cell voltage.

The objectives of these experiments are:

To illustrate that there is in the course of time a change in
the externally measured cell voltage, between the glass melt
and the AZS, due to the diffusion of cations of the melt into
the AZS (change in ~E) and the change in internal i~(t) by a
change in the size of the reduced layer.
To determine if the change in this external cell voltage
depends on the chemical composition of the molten glass.

At t=O the measured external cell voltage depends on the 'internal

iR drop' and the EAZSlmlPl' The EAZSlmlPl (15) is a function of the
standard Gibbs energy and for instance the 0 2- activity, both are
dependent on the glass composition.
In the case of the measurement of the external cell voltage the ~

is equal to the summation of RM + and ~. The electrical current in

the internal cell is i;.
The test set-up of figures 2.2.2 and 2.2.3 has been applied and
the external cell voltage between the platinum electrode in the
glass melt and the platinum ring at the outside of the crucible
was measured. The AZS 32 (ER 1681) crucible was filled with cullet
and heated up with 175 °C/h up till the test temperature. As soon
as the required temperature has been reached, the test starts.
After 90 and 160 hours, the glass melt has been renewed with a
fresh glass melt of the same composition. During the glass renewal
the crucible has been taken out of the test furnace and the glass
has been poured out. In the meantime fresh glass is taken out of
an other furnace with the same temperature as the test furnace.
The AZS test crucible is immediately refilled by the freshly
molten glass and placed in the furnace again. The duration of the
glass renewal procedure is about 1 minute. The glass melt has been
renewed to decrease the effect of the change in glass melt
composition (due to diffusion of cations and AZS corrosion
products) on the measurement of the external cell voltage.
Figure 2.2.4 shows the in this study measured time dependent
 33

figure 2.2.4 Measurements of the external cell voltage of three

different glass types at different temperatures as a
function of time.

1350 C figure 2'.2.4.8

External cell voltage mY.

800 - - - - - - - - - - - - - - - - - - .--- - - - - - - ----_."---'

800

~~---...~--
600

400 ~ ---"'++.
300·

200
1 glass renewed I
100

o ----.L.....-...___ _ L _

o 50 100 150 200 250 300

hrs.

-+- lithium glass -*-- 395 glas8 -(t- 354 glass

1425 C figure 2.2A.b

Ex ternal cell voltage mV.

800 -
-.----~-------.---.--------- --

600

400

300

200

100

oLI----'
o 60 100 150 200 260 300
hrs.

-+-lithium glass . ...,r,- 395 glass -~- 354 glas8

1500 C figure 2.2.4.c

External cell voltage mV.

800 - - - - _ . _ - - - - _... _ . _ - - - -

700

600

600

400

300

200

'00

tOO '50 200 260 300

hrs.

--. lithium glass --(0 395 0la88

 34

external voltages of three different glass types at different

temperatures.
The graphs show for 354 and/or 395 glass (see annex 1 for the
glass composition) a drop in the Ec in the course of time for the
three test temperatures. At higher temperatures the measured drop
in ~ is larger. After 90 hours the glass is renewed and the
measured Ec increases almost to the starting values in the tests
at 1425 and 1500 °c and even at a higher value than the starting Eo
value in the test at 1350 °C.
The new glass has the original composition without losses of
cations which have been diffused into the AZS and also without the
corrosion products of the AZS in it. This means that the Gibbs
energy difference between the glass melt and AZS increases again
due to the renewal of the glass melt. Thus the driving force of
the cation diffusion is increased by the renewal of the glass melt
and therefore the internal current of the internal cell will
increase, with a internal cell resistance practically equal just
before and after the glass renewal.

Experiments with 354 and 395 glasses at 1350 °C.

In the experiments with 354 and 395 glass at 1350 °c the Eo just
after glass renewal at 90 hours is even larger than the starting
value. The Ec measured after the glass renewal at 160 hours, is
larger than the Ec after 90 hours (see figure 2.2.4.a). This
implies that, in the time between glass renewal, the change in
Gibbs energy of the total practical cell ~E is larger than
the change in iiR; of the internal cell. The i;R; of the internal
cell is probably decreasing during the experiments of 354 and 395
glass at 1350 °c, although the R; is increasing due to the increase
of the size of the reduced layer. Therefore, if the proposed
mechanism is valid, the decrease in the i j seems to be relatively
larger than the increase of the R;. The overall effect is a small
decrease of the ijR; during the experiment at this temperature and
these glasses. This would explain the larger Eo values after each
glass melt renewal at 90 and 160 hours.
 35

Experiments with 354 and 395 glasses at 1425 and 1500 ·C.
In the experiments with the same glasses (354 and 395 glass) at
1425 and 1500 ·C there is a decrease in time for the measured
external cell voltage. Also the values at 90 and 160 hours, just
after glass renewal, are decreasing in the test duration (see
figure 2.2.4.b and 2.2.4.c). This implies that either the internal
ii~ of the internal cell in the duration of the experiment is
increasing or the chemical composition of the AZS, due to the
diffusion of the cation of the molten glass into the AZS, which
leads to composition changes even at the outer surface of the AZS
at the platinum ring electrode. This brings a reduction of the
Gibbs energy difference in the system and therefore in the
measured ~. The increase of the ij~ value of the internal cell in
the duration of the experiment is contradictory to the
observations at 1350 ·C with these glasses. The decrease of the
measured Ec , at 1425 and 1500 ·C in the experiments with 354 and
395 glass, is probably not caused by an increase in ii~. This
decrease of the Ec is probably caused by the decrease of the
concentration of cations in the glass melt by diffusion into the
AZS on one hand and the increase of the concentration of the
diffused cation in the AZS till the platinum electrode at the
outside on the other hand. The change in E, due to the decrease of
cations in the glass melt is illustrated by the increase of the ~

just after the renewal of the glass in figure 2.2.4. The only
small decrease in ~ after glass renewal at 90 hours compared to
the starting value of the E, and the larger drop of the ~ after
the glass renewal at 160 hours compared with the starting value of
the E, agrees very well with the measured change in chemical
composition of the AZS material at the outside of the crucible at
the platinum electrode. This will be presented in chapter 3.

Experiments with lithium glass.

For the lithium glass (see annex 1) the same argumentation can be
applied, with the remark that the mobility and the penetration
depth of the cation(s) has to be much larger as in the other
glasses to explain the measured E, values. Indeed, the larger
 36

mobility and penetration rate of the cations of the lithium glass

into the AZS will be demonstrated in chapter 3 and 4 of this
study.

The time dependency of the measured Ec and the strong influence of

temperature and the glass melt composition, means that the rate of
the internal reaction (5) probably depends on the velocity of
diffusion of the cations from the melt into the AZS refractory.
The resistance of the (internal) short-circuiting of the internal
cell ~ (10) can therefore not be decisive, otherwise the glass
melt composition would not have such a large influence on the Ec •
Indicating that probably,
~+ > ~ or (C Mo +. UMo +) < (C.-. U.-)

2.2.3.2. Shift of the location of oxygen formation.

Cell short-circuiting.

The objectives of these experiments are:

To determine whether by short-circuiting, in the test set-up of
figures 2.2.2 and 2.2.3, a part of the oxidation of the oxide
ions can be shifted from the AZS surface to the platinum-to-melt
surface.
- The measurement of the electric (external) current, in order to
obtain an estimation of the resistance ratio for electron
transfer between the non reduced part and the reduced part of
the AZS.
Besides the short-circuiting of the cell, the test circumstances
were the same as in the previous tests described in 2.2.3.1.
In a circuit in which a platinum electrode in the melt, is short-
circuited with the AZS, oxygen bubbles are formed at the surface
of this electrode. This means that part of the oxidation of oxide
ions has shifted from the AZS surface to the platinum-to-melt
interface (14) with 0 < x < 1. Besides, electrons migrate from the
platinum electrode to the reduction zone under the AZS surface,
via the external circuit and the non reduced part of the
 37

refractory. The electric current measured on a short-circuited

cell shows to be sensible for small differences in circumstances,
figure 2.2.5 gives a good example of the tendency and order of
magnitude of the measured current.

figure 2.2.5 Measured external current of a short-circuiting cell.

Test with AZS 32 (ER 1681).

Short-circuiting cell
354 glass, 1425 C
rnA
2.5,----------------------------,2.5

2 2

1.5 1.5

glass renewed
0.5

oL.-----'-------"--------'-------L-------'-----O
o 50 100 150 200 250 300
hrs.

-+- current

The steep rises in electric current during the two tests, one
after 90 and one after 161 hours, are due to the renewal of the
glass, which suddenly increases the Gibbs free energy between
glass melt and AZS again. This is because the new glass has the
original composition without loss of cation, which has been
diffused into the AZS. The new glass also does not contain
corrosion products of the AZS.
The low initial current, at t=o till t=50 hours, in the test is
probably caused by the poor electrical contact between the
 38

platinum outer ring and the initially "dry" AZS. After the test
temperature is reached, the AZS starts to expel glassy phase and
the electrical contact between the surface of the AZS and the
platinum ring improves. After about 50 until 90 hours the measured
electric external current increases. In the middle of the duration
of the experiment, t=90 till t=161 hours, the measured electric
current is more or less stable. At the end of the test, t= 161
till t= 260 hours, the measured electric current is decreasing.
To explain the measured current in figure 2.2.5, two time
dependent effects have to be considered, the thickness of the
reduced layer and the driving force.

On the time scale, the thickness of the reduced layer increases at

the expense of the original (non reduced) refractory layer
thickness. Figure 2.2.4.b gives the external voltage of a not
short circuiting cell, at 1425 °c using 354 glass. The external
cell voltage (as a measure for the driving force) initially is
about 700 mV and finally, after 260 hours, around 400 mY.
The increase of external current between 50 and 90 hours of the
test duration is probably caused by the increase of the reduced
layer and therefore the decrease of the non reduced layer.
In the middle part of the experiment the possible increase of
measured external current due to the increase of the reduced layer
has been compensated by the decrease of driving force. In the last
part of the test duration the decrease of driving force is larger
than the possible increase of external current due to the
decreasing layer thickness of the non reduced refractory.

Electrolysis by external voltage U. at the outer surface of the

AZS crucible.

The objectives of these experiments are:

- To investigate the possibility of suppressing bubble formation
on the interface glass melt/AZS.
- To examine the influence of the external voltage on the
diffusion of the cations.
 39

- To determine the relation of external voltage on the electric

current.

Test set-up
The set-up is the same as for the previous tests (see figures
2.2.2 and 2.2.3). The only difference in this test is the
implementation of a fixed voltage between the platinum ring
surrounding the AZS crucible (negative pole) and the platinum
probe in the glass.
For the electrical network see figure 2.2.3.
Test conditions:
applied voltage 0; 1; 2.5; 5 or 39.5 V
test period 48 hours (one 72-hour test)
temperature 1425°C
glass type 354 glass
probe area 800 and 20 mm2
crucible diameter inside: 70 mm; outside: 89 rom
material AZS 32 (ER 1681)
The parameters that have been analyzed or measured, are the
electric current, the cation diffusion (concentration profiles) in
the glassy phase of the AZS after the exposure time and the bubble
formation at the AZS-to-glass interface.

A) Electric current.

The measured current running through the cell varied (see figure
2.2.6.a). At 39.5, 5 and 2.5 V the measured amperage drops at
first; at 1 V it increases slightly at first. After about 48 hours
the electric current is virtually identical for the four
differences in potential, and remains stable for extended periods
(figure 2.2.6 b).
The current density depends on the probe area. The steady-state
amperage with the 800 mm2 probe was about 28 rnA and that of the
20 mm2 probe about 12 rnA (see figure 2.2.6 a and 2.2.6 c). This
indicates that the electrical resistance at the interface
platinum-glass is relatively large.
 40

figure 2.2.6 Measurement of the current through the external

circuit, applying different external voltage's.

Electrolysis
current amperage on time scale
amperage (mA) figure 2.2.6.a
350 r--=--------:::-------------------~ 350

300 300

250 250

200 200

150 150

100 100

50 50

0 0
0 20 40 60 80
time (hours)

39,5 V--+-5
-H-- V -«--- 2,5 V --e- 1 V -- 1 V
applied external voltage
1425 C. 354 glass
800 mm2 electrode area

Electrolysis
current amperage on time scale
amperage (mA) figure 2.2.6.b
200 ~-=-----=~------------------___, 200

150 150

100 100

50 50

o "0
o ''-- ---I... -'--- -----' ---'--
c

-----.J 0
o 10 20 30 40 50
time (hours)

--- 2.5 V ......... 39.5 V

applied external voltage
1425 C. 354 glass
20 mm2 electrode area
 41

B) Diffusion of cations into the glassy phase of the AZS.

To study the effect of the voltage on the diffusion processes,

special attention has been paid to the concentration profiles of
NazO, KzO, BaO and Al z0 3 in the AZS after an experiment. Annex 2,
table 1 gives the measured values.
The method which has been applied for the determination of the
concentration profiles will be given in the next chapter.
For the present tests, the concentration profiles in the glassy
phase in the AZS have been determined as closely as possible to
the platinum ring position surrounding the outer part of the
crucible (measuring concentration profiles at the same vertical
positions). The value of the applied external voltage, appears to
have hardly any systematic effect on the concentration profiles of
KzO and BaO. See figure 2.2.7, 2.2.8 and annex 2, table 1 for the
measurements using a 20 mmz glass melt electrode area.
However, the effect of the external voltage on the profile of NazO
at the outside of the crucible is considerable for 354 glass. The
tests with an external negative potential between the AZS
electrode and glass electrode showed a strong accumulation in NazO
concentration deep in the AZS at the position close to the
platinum ring surrounding the outer part of the crucible. See
figure 2.2.9.
The difference in potential (extra negative charge at the crucible
electrode) has brought about additional diffusion of sodium from
the glass melt into the glassy phase of the AZS towards the
negative pole at the outer surface of the crucible. This
accumulated additional Na+ diffusion increased for higher
potentials. The AIZO] concentration in the glassy phase of the AZS
increases along with the NazO concentration. See figure 2.2.10.
This is probably due to the increase in solubility of AIZO] (from
corundum) in the glassy phases with an increased NazO
concentration.
The mobility of cations in glass depends on the ion radius and ion
charge. The fact that the sodium ion has the highest velocity
could, therefore, be accounted for, from the combination of the
 42

figure 2.2.7 and 2.2.8 Concentration profiles of K20 and BaO after
applying different external voltages.

figure 2.2.7
Electrolysis
K20 diffusion
K20 weight %
8.----=------------.-------------,

glass
8

o ~_---'- _ __'___ ___'____ __'_____L______'_ _____'__---L_ __'_____ _'__________''________'__ __'___---l

-8 -7 -6 -5 -4 -3 -2 -1 0 2 3 4 5 6
mm

- 0 V---+-1 V ---e--- 2,5 V --;<--- 5 V --+- 39.5 V

applied external voltage
1425 C. 354 glass, after 48 hours
800 mm2 electrode area

figure 2.2.8
Electrolysis
BaO diffusion
BaO weight'll.
12

10
AZS glass
8

0
-8 -7 -6 -5 -4 -3 -2 -1 0 2 3 4 5 8
mm

- 0 V---+-1 V ---e--- 2,5 V ---><-- 5 V --+- 39,5 V

applied external voltage
1425 C. 354 glass, after 48 hours
800 mm2 electrode area
 43

figure 2.2.9 and 2.2.10 Concentration profiles of Na20 and Al 20 3

after applying different external
voltages.

figure 2.2.9
Electrolysis
Na20 diffusion

16 Na20 weigh_t_% .--- ---,

14
AZS glass
12

10

2
oL--1 -'--------'--
-6 -7 -6 -5 -4 -3 -2 -1 0 2 3 4 5 6
mm

- 0 V-+--l V ---- 2,5 V ---- 5 V -+- 39,5 V

applied external voltage
1425 C, 354 glass, after 48 hours
800 mm2 electrode area

figure 2.2.10
Electrolysis
AI203 diffusion
AI203 weight %
40

glass
30

20

10

o
-6 -7 -6 -5 -4 -3 -2 -1 o 2 3 4 5 6
mm

- 0 V-+--l V ---- 2,5 V ---><- 5 V -+- 39,5 V

applied external voltage
1425 C, 354 glass, after 48 hours
800 mm2 electrode area
 44

ion radius (Na+ 0.97; K+ 1.33; and Ba 2 + 1.34 A) and the ion
charge. The double charge of an ion reduces the mobility of the
ion dramatically compared to an univalent ion of the same radius
(Lit. 18). The effect which the electrical field of an ion has on
the fundamental relations in oxide systems are described by A.
Dietzel (Lit. 19). The effect of the electric potential on the
velocity of diffusion of Na+, which is far stronger than in the
case of the other cations, has been determined on earlier
occasions. Doremus (Lit. 15) has determined the mobilities of
alkali cations in vitreous silica under the influence of an
electric potential at lower temperatures (max. 380°C).
The mobility of Na+ was a factor of 1000 higher than that of K+.
That of Li+ (ion radius 0.68 A), however, was far lower (about a
factor of 10) than that of Na+. The latter fact, therefore, is not
in agreement with the assumption that small ion radius, show a
higher mobility under the influence of an electric potential.
If Doremus's estimate of the relative mobility of the various
cations also applies to the system and temperature of these
investigations, the mobilities of the other cations are too low
for the determination of the difference in concentration profiles
of these cations between the presence and the absence of an
external potential using this coarse test method. Also the
platinum ring with a wire diameter of 1 mm surrounding the outer
crucible surface is not an optimum condition for a quantitative
measurement, because the increase of the Na 20 concentration was
half cylindrical around this very local platinum-AZS contact.
Theoretically, a platinum AZS ribbon along the total height at the
outer wall would be a large improvement for more quantitative
tests, but it is practically impossible to establish a good
electrical contact between the total ribbon surface and the
crucible wall during the experiment.

C) Bubble formation at AZS/glass melt interface.

The appiication of an external voltage (negative pole at the outer

wall) shows a reduction in the number of bubbles generated in the
 45

figure 2.2.11
Electrolysis
number of bubbles from AZS

bubbles/cm3
50,---------------------------------,

20

10 +

+
o l-._---"--_"---'---L'i'--l'--l'i'-L'_ _--"--------"_---.JLi-.LLLLL_ _L----"---'---'-i--'--'--U
0.1 10 100
Volts

+ electrode 800 mm2 o electrode 20 mm2

1425 C. 354 glass. after 48 hours

AZS/glass melt interface.

An increase in potential results in a drop in the number of
bubbles, counted after 48 hours, at this interface. See figure
2.2.11. There is a linear relationship between the number of
bubbles and the logarithmic value of the voltage. Although the
external current, after 48 hours, is independent of the external
voltage in this test set-up, the external current depends on the
used electrode in the glass melt. Figure 2.2.11 also shows the
smaller drop in the amount of bubbles at the AZS interface at the
end of the tests using the 20 mm2 electrode compared to the 800 mm2
electrode. The measured quantitative relationship probably depends
on the applied test set-up, specially the used electrodes in the
glass melt and at the AZS crucible outer wall. If an external
voltage with a positive pole at the outer wall of the crucible
would be applied, according to the theory, the number of bubbles
generated at the interface glass melt/AZS would be increased. The
 46

internal reduction rate however would be reduced. In the case of a

positive pole at the outer wall of AZS crucible, at a high
external voltage, the cation migration can even change in the
direction from the glassy phase of the AZS into the glass melt.

2.2.4 Discussion and conclusions.

External cell voltage:

In this study it has been measured and shown that the rate of the
decrease of the external cell voltage, and therefore a change in
Gibbs energy (8), during a test depends on the chemical
composition of the attacking glass melt.
This implies that the rate of the internal reaction probably
depends on the penetration of the different cations (depending on
the glass composition) in the glassy phase of the AZS.

Cell short-circuiting:
In a short circuited external cell, some oxygen bubble formation
on the platinum electrode in the glass melt has been observed
which means that a part of the oxidation of oxide ions has shifted
from the AZS/glass melt interface to the platinum-to-glass melt
surface.

Electrolysis by external voltage:

with an external voltage, with the AZS crucible outer surface as a
negative pole, it is possible to reduce the amount of oxygen
bubbles generated in the AZS-to-glass melt interface.
These measurements and observations also confirm the proposed
mechanism.
In an industrial process it will be difficult to apply an external
voltage for bubble reduction because then, at the positive pole
(in the melt) oxygen bubbles will be formed. The outside of the
AZS is relatively cold with a high electrical resistance at these
low temperatures. Sodium cations accumulate in the AZS which
increases the corrosion and affects the mechanical properties of
the AZS material.
 47

In the electrolysis by an external voltage, with the AZS outer

surface as negative pole, the following phenomena have been
examined:
1. The electrode area for the electrode in the melt as well as
probably (not investigated) also for the platinum wire at the
AZS surface affects the external measured current at a specific
voltage.
2. After some time, at different externally applied voltages, the
external current tends to the same steady-state level, only
dependent on the electrode area.
3. The number of bubbles at the AZS/glass melt interface, after 48
hours, depends on the applied external voltage.
4. The initial current density of an electrode surface depends on
the applied voltage.
5. The diffusion rate of Na+ increases markedly with an increase
in voltage. This increase, like the increase in external
current, is remarkably higher for the 800 mm2 glass melt
electrode than for the 20 mm2 one.
6. At a specific voltage level, the glass melt electrode area also
affects the bubble suppression.

These six items justify the conclusion that the resistance of the
external cell with the measured current densities depends on time
and the area of the platinum electrodes.
The ratio of the applied fixed external voltage and the measured
external current values (Vem~l/i'~) gives the resistance of the
test set-up. In other words the sum of the resistances of the
platinum wires, platinum electrode to glass melt, glass melt
itself, glass melt to AZS, AZS to platinum ring, the non reduced
AZS and the reduced AZS. During electrolysis the resistance of the
reduced layer, according to Baucke and Roeth (Lit. 7), can
formally be treated as, (RM+.~/(RM++~)).
The so calculated resistances are for instance dependent on time,
external voltage and surface area of the platinum glass melt
electrode. In figure 2.2.12 the ratio of the calculated total
resistances is given as a function of time for the two used
 48

Ratio of total resistance between the figure 2.2.12

glass electrodes of 20 and 800 mm2,
when external voltage is applied.

resistance ratio
3.5 ~=---=-':'=':"':"":=--=---=------------------,

3
2.5

1.5

0.5
o L- ---'-- L- ----'--_~ l_ ~

o 10 20 30 40 50
time (hours)

~- 2.5 V -+- 39.5 V

external voltage

platinum electrodes (resistance ratio(t)

{ (Ve_ / i' .. ) 2Onun2/ (Vextenal / i' eJ gOO mm2} (t) at two different externally
applied voltages (2.5 and 39.5 V).
The ratio between the measured resistance, for the case of the 20
mm2 electrode divided by the case of the 800 mm2 electrode, is more
or less constant in time, which implies that the main resistance,
from the beginning, lies here at the interface platinum electrode/
glass melt where oxygen bubbles are formed. The decrease of the
electric current during the experiments when an external voltage
is applied, means an increase of the electric resistance in time
mainly at the interface platinum electrode/glass melt. An increase
of resistance of a platinum electrode in a glass melt is a well
known phenomenon. Normally this is due to a electrical double
layer which establishes itself very fast (milliseconds) (Lit. 14).
In this study it takes up to two days in order to obtain a
constant current level and thus a constant resistance. Such a time
 49

scale, points in the direction of a diffusion limited process.

The behaviour of equal electrical current for different potentials
resembles that of a complete concentration polarization (Lit. 14).
At the platinum electrode in the glass melt oxygen bubbles are
formed, and are released from the glass melt. It is possible that
the transport (diffusion) of 0 2 - ions in the glass melt to the
platinum electrode is the cause for the increase of the resistance
and the 'steady state' level of the total resistance of the
measured electrical loop after about 48 hours. In other words, the
increase of the resistance possibly could be caused by the
formation of a 0 2- diffusion layer.
A larger surface of the platinum glass melt electrode and the
already mentioned platinum cylinder instead of the platinum wire
at the outer surface of the AZS crucible would improve the
quantitative accuracy of the measurements.
The external voltage level affects the bubble suppression at the
interface glass melt/AZS (see figure 2.2.11). The higher the
external voltage with the negative pole at the outer wall of the
crucible the lower the number of bubbles generated at the
interface glass melt/AZS. A low number of bubbles at the interface
glass melt/AZS means that the amount of electrons i~ formed by
2
oxidation of 0 - to O2 is decreased at the interface glass
melt/AZS. This is to be expected because the electrons (i~) from
the glass melt/AZS interface now have to move in the direction of
the negative pole of the implied external voltage.
The electric current due to the migration of cations is equal to
the summation of the current of electrons in the internal cell and
the non reduced AZS.

The ultimate externally measured amperage (i'~) in the final test

phase (after 48 hours) is the same at the various different
external voltages. This means that after 48 hours, at a higher
voltage with less bubble formation (less i~) at the interface
glass melt/AZS, the cation diffusion into the AZS has to be
 50

smaller.
The statement that at a higher externally applied voltage, after
48 hours, the amount of cation diffusion into the AZS becomes
smaller as the current is equal for all tests with the same
electrode, looks strange at first. Above it has been explained
that the main part of the electrical resistance, is at and close
to the interface platinum electrode/glass melt. The potential drop
across the AZS in the tests caused by the externally applied
voltage, after 48 hours, is not so dominant anymore. Under these
conditions the differences in the chemical partial Gibbs energy of
the cations in the glass melt and the glassy phase of the AZS
become important again as driving force, for the diffusion rate of
cations (i Mn +). In figure 2.2.9 it is shown that the tests with a
high externally applied voltage lead to a high concentration of
sodium oxide in the glassy phase of the AZS.
This high concentration of sodium oxide in the glassy phase of the
AZS is accumulated in the first 24 hours of the test duration,
when the measured external current has been large compared with
the tests with no or small externally applied voltages. Therefore,
after 48 hours, the rate of cation diffusion in the tests with
high externally applied voltage has decreased considerably and is
smaller than in the tests with low externally applied voltages.
The decrease in rate of cation diffusion is caused by the decrease
in difference if Gibbs energy between the AZS and the molten
glass.
In this study, the objective has been the validation of the
proposed mechanism for fused cast AZS in contact with molten
glass. A more detailed study with different glass compositions
(e.g. sodium free glass) of the effects of external electric
potential on the cation diffusion and suppression of bubble
formation at the interface glass melt/AZS could increase the
insight of the mechanism. Also the hypothesis of the formation of
a diffusion layer for 0 2- at the platinum electrode in the glass
melt as the main electrical resistance if a external voltage is
applied, needs more intensive study.
Nevertheless, the objective of the validation of the mechanism in
 51

which the diffusion of cations into the glassy phase of the fused
cast AZS is controlling the formation of oxygen bubbles at the
interface glass melt/fused cast AZS has been met.
 52

Literature references chapter 2.

[ 1] F.W. Kraemer;
Analysis of gases evolved by AZS refractories and by
refractory/glass melt reactions. Techniques and results.
Contribution to the bubble forming mechanism of AZS
material.
Glastech. Ber. 65 (1992) p. 93-98.
[ 2] E.L. Swarts;
Bubble generation at glass/refractory interfaces. A review
of fundamental mechanisms and practical considerations.
Glastech. Ber. 65 (1992) p. 87-92.
[ 3] M. Dunkl;
Studies on the glassy and reaction phases given off by fused
cast AZS blocks and their effects on glass quality.
Glastech. Ber. 62 (1989) p. 389-395.
[ 4] D. Walrod;
A study of the driving force behind AZS glass phase
exudation.
Ceram. Eng. Sci. Proc. 10 (1989) p. 338-347.
[ 5] A.G. Bossard, E.R. Begley;
Refractory blistering in glass.
Symposium on defects in glass. An. Meet. ICG.
Tokyo 1966 p. 69-81.
[ 6] L. Leger, M. Boffe, E. Plumat;
Electrochemical phenomenon at the glass-refractory material
interface.
Glass Technology 1 (1960) P 174-179.
[ 7] F.G.K. Baucke, G. Roeth;
Electrochemical mechanism of the oxygen bubble formation at
the interface between oxidic melts and zirconium silicate
refractories.
Glastech. Ber. 61 (1988) p. 109-118.
[ 8] W. Trier;
Glasschmelzoefen, Konstruktion und Betriebsverhalten.
springer Verlag 1984 p. 64
ISBN 3-540-12494-2
 53

[ 9] H. Meyer, H. Poehnitzsch;
Ueber die Ursache des Glasaustritts und der Blasenbildung an
schmelzgegossenen steinen bei hohen Temperaturen.
Glastech. Ber. 38 (1965) p. 393-397.
[10] o. Schmid;
Ueber die Glasphase in schmelzfluessig gegossenen
Aluminiumoxyd-zirkonoxydsteinen.
Glastech. Ber. 38 (1965) P 200-206.
[11] M. Dunkl;
Investigation of the liberation of glassy phase from fused
cast AZS materials.
Glastech. Ber. 63K (1990) p. 370-380.
[12] P. Ratto;
What can we expect from fused cast refractories.
Glass machinery 1991 p. 62-64.
[13] G. Meden, T. van der Pas;
Invloed van het uitstoken van ZAC 1681 op de mate van
belvorming.
OC 81/350.
[14] A.J. Bard, L.R. Faulkner;
Electrochemical methods; Fundaments and applications.
John Wiley & Sons, New York 1980.
[15] R.H. Doremus;
Electrical conductivity and electrolysis of alkali ions in
silica glass.
Phys. and chem. of glasses, 10 (1969), p. 28-33.
[16] F. Daniels, R.A. Alberty;
Physical chemistry, Chapter 7.
John Wiley & Sons. New York 1966.
[17] G.H. Frischat;
Ionic diffusion in oxide glasses.
Trans Tech S.A.
[18] H.H. Blau;
Fourth Int. Congr. on Glass, Paper VI 6.
Paris, 1956.
 54

[19] A. Dietzel;
Die Kationenfeldstarken und ihre Beziehungen zu
Entglasungsvorgangen, zur Verbindungsbildung und zu den
Schrnelzpunten von Silicaten.
z. fUr Elektrochern. 48 (1942) p. 9-23.
 55

3. cation diffusion into fused cast AZS.

3.1 Introduction.

When a glass melt is brought into contact with AZS, due to a

difference in chemical potential, a transport process of cations
will take place, to come to chemical equilibrium. simultaneously
with the transport of positive load, due to the transport of
cations in the AZS, electrons are transported to the reduction
front in the interior of the AZS and redox reactions take place at
the refractory/glass melt interface (oxygen formation) and at the
reduction front in the interior of the AZS (reduction of
polyvalent electrons). The combined transport process of cations
and electrons determines the oxygen bubble formation at the
refractory/glass melt interface. The underlying mechanism has been
described in chapter 2 and this is the main mechanism responsible
for the formation of bubbles and knots at the interface of fused
cast AZS and molten glass in industrial operation.
In this chapter the experiments with fused cast AZS 32/33 (ER
1681) crucibles and different glasses are described. In the first
part of the chapter the concentration profiles of sodium,
potassium, barium and strontium in the melt at the interface with
AZS and the glassy phase of the AZS are shown. The chemical
composition of the interface glass melt/fused cast AZS should
correspond with the original chemical composition of knots and
will be used in chapter 6 'Knot formation mechanism and
characterization' for the identification of knots. The
concentration profiles in the glassy phase of the AZS give
information about the penetration depth of the cations. This is
important for the second part of this chapter. In the second part
of this chapter the 'total or overall' diffusion rate of cations
is derived. For a straightforward determination of the diffusivity
the penetration depth of the cations should not have the tendency
to exceed the crucible wall thickness. The overall diffusion rate
of a cation in a crucible experiment is determined by measuring
the change in chemical composition of the glass melt as a function
 56

of time. Here, the rate of diffusion will be expressed in so

called electron equivalent per hour (e.e.h.) which is the positive
charge transported by the cation in the fused cast AZS, determined
in a standard test set-up. The e.e.h. is the result of a
combination of effects like AZS surface area, mobility of the
cation, difference in partial Gibbs energy of the cation in the
melt and in the glassy phase of the AZS and the time the AZS has
been exposed to the glass melt.
The e.e.h. is therefore only a relative value for the diffusion
rate. In chapter 4, a model will be presented which determines the
influence of the temperature and chemical composition of the glass
melt on the formation of glass defects from the fused cast AZS
using the measured e.e.h. values.

3.2 Experimental procedures.

Test data:

crucible ER 1681 (Fused cast AZS)

inner diameter 70.5 mm
outer diameter 89.0 mm
height 75 mm
depth 52-55 mm
weight about 950 g
glass types 395 glass: table 3.1
354 glass: table 3.1
lithium glass: table 3.1
glass quantity about 360 9 (cullet)
temperatures 1350; 1425; 1500 0 C
electrical heated furnace (Naber)
test periods: a) 100 hours, including at the end a
temperature alternation (last nine
hours): four times a cycle of one hour
at standard temperature +25°C and one
hour at standard temperature -25°C,
followed by one hour at standard
 57

temperature (described as: "100 hrs.

alt. temp." in figures).
b) 100 hours with forced convection in melt,
at a rotation speed of w=0.2 and
2.0 cycles min-1. using a rod shaped
stirrer with diameter of 21 rom, stirrer
height 13.5 mm above crucible bottom.
(described as: "100 hrs. w=0.2" or "100
hrs. w=2.0" in figures).
c) 260 hours:glass renewal after
about 90 and after 160 hours
d) 600 hours:glass renewal after about 90,
163, 258, 331, 426 and 499 hours.

Note: The experiments with temperature alternation and forced

convection (increased corrosion) have been executed to determine
the influence of these parameters on the bubble formation (see
chapter 5).
Neither the temperature alternation nor the application of
shear stress (owing to forced convection) had any appreciable
effect on either the diffusion profile or the concentration
profile in the AZS. This implies that these tests are indicative
of the reproducibility of the measuring method.
This also implies that the diffusion in the glass melt is not the
rate determining step of the electrochemical process but the
diffusion of the cations in the glassy phase in the fused cast"
AZS.

The cation diffusion rate has been determined:

A. by analysis of the profiles of the concentration of the cations
in the glass and the glassy phase in the AZS.
B. by analysis of the composition of the glass which was renewed
during the 260 and 600 hour tests.
 58

3.2.1. Profiles of the concentration of the cations in the glass

melt and the glassy phase of the AZS.

3.2.1.1 Measurement method.

The crucible has been filled with cullet and has been heated, with
a rate of 175 °C/h, up to the test temperature. The test time
starts as the test temperature has been reached.
After completion of the experimental melting time, the crucible
has been taken out of the furnace and cooled during 5 minutes at
room temperature. The crucible has been subsequently put into an
annealing furnace at 520°C and cooled down slowly overnight. The
cracks in the glass were filled with 'Canada Balsam', of which the
volatile part has been evaporated at 130°C. From the crucible a
slice has been cut of about 10 mm thickness. A sample with a
diameter of 20 mm has been drilled from this slice, with the
interface AZS-glass in the middle of the sample. The sample was
polished for measurements using a scanning electron microscope
(SEM) and energy dispersive X-ray analysis (EDX) system.
The SEM is a Philips 505 and the EDX, an Edax PV9900.
Every spot measurement (very localised EDX measurement) was done
under identical conditions:
time 100 live (real measurement time) seconds
magnification 5200 x
selected area 6.3 x 4.6 micrometer per spot measurement
current 0.70 nA
voltage 25 kV
sample position: eucentric (34.5 mm)
The fluorescence peaks in the measurements of the intensities can
be seen in figure 3.2.1.
The measured intensity is calculated compared to a standard sample
using background correction. The local concentration of the
different elements is calculated as oxides in weight% and has been
normalised to 100%.
The chemical composition of the glass and of the glassy phase of
the AZS matrix has been measured. The dividing line (zero in the
 59

figure 3.2.1 Fluorescence peaks in the EDX measurements of the

intensities.

i SiV.CY.
i \
i I
ntensity
I\
r I K ~«:C:

1
1 tt
NaK~
n
1\
AIK", I III
{I J\ f \ fI
1\ ~
l'aL""

I \ "'~K",\t \\
J\
ZrL", I \l/M,~o~" CaK"
J \ Sr~<.;..:
. . 1 \r''Y
--- I ,
~-
I
----....
I
::1
L.
~
I I
V I I DII!:::;:::,.,..,...",,"=:.....:::l<;;;;ll;'..........
1,00 2.00 3,00 4,00 5,00 14,00

KeV

figures) between glass and AZS is formed by the last Zr02 nodules.
The measurements started at a distance of 6000 micrometer from
this dividing line in the glass. By definition, a negative
distance means, the distance from the dividing line and measured
spot in the glassy phase of the AZS. The standard procedure was
always two spot measurements on 6000, 4000, 2000, 1500, 1000, 750,
500, 400, 300, 200, 150, 100, 50, 0, -50, -100, -150, -200, -
300, -500, -1000, -2000, -4000, -6000 micrometers. In this way the
spot measurements started at the glass side (positive value) going
to the glassy phase (negative value) of the fused cast AZS. Close
to the interface glass/AZS, the number of spot measurements were
increased.
 60

3.2.1.2 Results of measured concentration profiles.

The concentration profiles of the most important cations, which

can be measured with EDX in the glasses from annex 1, are Na 20,
K20, BaO and SrO. Here the measured concentration profiles of
these oxides will be presented in detail.

The general shape of the concentration profile of Na20 from the

glass side into the fused cast AZS, (in the figures from right to
left) is a slight concentration decrease, just before the first
Zr02 nodules are reached and than even an increase to a maximum
concentration higher than in the used glasses.
Table 2 in annex 2 shows the measured values. An example of the
reproducibility is presented in figure 3.2.2. An example for the
effects of the progress of the concentration profile during the
exposure time is presented in figure 3.2.3. The penetration depth
of the diffusion into the AZS shows the expected strong increase
with the duration of the test period.
The maximum Na20 concentration in the AZS generally increases
slightly in the course of time. Owing to the increase in diffusion
depth going with longer periods the location of the maximum
concentration is shifted ever deeper into the AZS.
The effect of the temperature is presented in figure 3.2.4. At
1350 0 C (the lowest test temperature) the maximum concentration is
generally highest and the penetration depth obviously smallest.
The effect of the different glass types is very obvious, for
instance given by figure 3.2.5. An increase in the concentration
of Na 20 in the molten glass yields an increase in concentration of
Na 20 in the glassy phase of the AZS. The sodium from the lithium
containing glass has an exceptionally large diffusion depth
compared to 354 glass and 395 glass.
The crucible wall was about 9 mm thick. At 1350 DC, after only 100
hours the penetration depth of Na 20 was markedly less than this 9
rom.
 61

figure 3.2.2
Na20 diffusion in AZS
354 glass, 1350 C
Na20 weight"
10,.-----------------,---------------,

AZS glaaa
8

O~-----'------'-------'---'--------'------'-----'-------'-----J

-10 -8 -6 -4 -2 o 2 4 6 8
mm
- 100 hr. -+- 100 hr. alt. temp -4- 100 hr. wo O.2 --a-- 100 hr. wo 2.0

figure 3.2.3
Na20 diffusion in AZS
395 glass, 1350 C
N.20 weiGht 'It
10,-----------------,---------------,

AZS glaaa
8

o '------~-----'-----'-------'--------<-----'------'-----'-------

-10 -8 -6 -4 -2 o 2 4 6 8
mm

- 100 hr. --+--- 260 hr. 2x renewed --+- 600 hr. 6x renewed
 62

Na20 diffusion in AZS figure 3.2.4

354 glass. 100 hr.
N.20 . .tght '4
10,--------------,------------,

AZS gla"
8

O'------'-----'--------L---'--------'----'----------'-----'----'
-10 -8 -6 -4 -2 o 2 4 6 8
mm

--+- mean 1350 C. -A-- mean 1425 C -H- mean 1500 C

figure 3.2.5
Na20 diffusion in AZS
1425 C. 260 hr.
Na20 .elght ..
10,--------------,---------------,

AZS glae,
8

.4

0'--------'----'----'-----'------'---'-----'------'------'
-10 -8 -6 -4 -2 o 2 4 6 8
mm

- 354 glass --+- 395 glass --+- Lithium glass

2>< renewed
 63

The general shape of the concentration profile of K20, going from

the glass into the fused cast AZS, shows an increase just before
the first Zr02 nodules are reached. The maximum of the first peak
(going from the melt into the AZS) is close to the position where
the first Zr02 nodules occur. After this first maximum, the K20
concentration decreases with about the same angle to a minimum and
increases again to a new maximum.
Table 3 in annex 2 shows the measured values. An example of the
reproducibility is presented in figure 3.2.6. An example for the
K20 concentration profile during the progress of time is presented
.in figure 3.2.7.
As with Na20, the penetration depth into the AZS shows the
expected strong increase with the duration of the test period. The
maximum K20 concentration in the AZS increases as the test lasts
longer. Owing to the increase in the penetration depth, after
longer periods, the location of the maximum K20 concentration is
shifted ever deeper into the AZS.
The effect of the temperature is shown for example in figure
3.2.8. At a low temperature the maximum concentration is generally
highest and the penetration depth smallest (the effect of the
temperature on the penetration depths is obvious in the
measurements after 260 hrs, see table 3 annex 2). In the glass-to-
AZS interface the K20 concentration goes up as the temperature·
goes down.
The effect of the use of different glass types is very obvious,
for example demonstrated by figure 3.2.9. An increase in
concentration of K20 in the glass yields an increase in
concentration in the glassy phase of the AZS. The K20 from the
lithium containing glass has an exceptionally large penetration
depth compared to 354 glass and 395 glass.The crucible wall was
about 9 mm thick. At 1350 0 C and after only 100 hours, the
penetration depth of K20 was markedly less than this 9 mm for all
cases.
 64

K20 diffusion in AZS figure 3.2.6

354 glass, 1425 C
K20 weight ..
12,-------------------.-----------_

AZS glaea
10

2L-----"-----L------'-----'-----'-_--'_ _---'---_ _-'-----_ _

-10 -8 -6 -4 -2 o 2 4 6 8
mm
260 hr. 2x renewed -+- 260 hr. 2x renewed
- - 260 hr. 2x renewed -&- 260 hr. 2x renewed

figure 3.2.7
K20 diffusion in AZS
395 glass, 1425 C
K20 weight '10
12 ,--------k----=-'~-------.-----------_

10

AZS gla08
4

2'---------'-------'-------'-------'------'-----'-----'------'-----~

-10 -8 -6 -4 -2 o 2 4 6 8
mm

- 100 hr. -+- 260 hr. 2x renewed ----- 600 hr. 6x renewed
 65

figure 3.2.8
K20 diffusion in AZS
354 glass, 100 hr.
K20 weight ..
12,-----------------.----------------,

AZS glasa
10

2 '--_----'--_ _--L-_ _ ---'~ _ ___'___ _l -______'__ ____'___ _L__----.J

-10 -8 -6 -4 -2 o 2 4 6 8
mm

- - mean 1350 C ---A-- mean 1425 C ~ mean 1500 C

figure 3.2.9
K20 diffusion in AZS
1425 C. 260 hr.
K20 weight ..
12,-----------"""""".---------,.-------------,

10

8
AZS gla••

2'------'--_ _L__----'--_ _L__----'--_ _L-_----'--_ _-L-_----.J

-10 -8 -6 -4 -2 o 2 4 6 8
mm

- 354 glass --+- 395 glass - - Lithium glass

2>< renewed
 66

The general shape of the concentration profile of BaO going from

the glass side into the fused cast AZS, shows a decrease starting
before the first Zr02 nodules of the AZS are found. The minimum
lies close to the first Zr02 nodules. From that point on the BaO
content increases to a maximum, and then drops relatively steep in
the direction of the AZS interior.
Table 4 in annex 2 shows the measured values. An example of the
reproducibility is presented in figure 3.2.10. An example of the
effects of the progress of time is presented in figure 3.2.11.
The penetration depth into the AZS increases along with the
increase in the test duration. The maximum BaO concentration in
the AZS remains unchanged (within the measuring inaccuracy
limits) •
The effect of the temperature is shown for example in figure
3.2.12. Both BaO concentration and the penetration depth into the
AZS go down as the temperature decreases.
The effect of the use of different glass types is very obvious,
illustrated by figure 3.2.13. An increase in the BaO concentration
in the molten glass (lithium free glass), yields an increase in
the concentration in the AZS (difference between 354 glass and 395
glass).
At 1350 °C (figure 3.2.14) and slightly less pronounced, at 1425°C
(figure 3.2.13) the BaO concentration in the case of lithium
glass, is higher in the glassy phase of the AZS compared to 395
glass with almost the same amount of BaO.
Again, the penetration depth of BaO from the lithium containing
glass is exceptionally large compared to the situations measured
for the 354 glass and 395 glass.
Only during the test with lithium glass after 260 hours at 1500 o C,
the penetration depth of BaO has the tendency to exceed the
crucible wall thickness of 9 mm (annex 2, table 4).
 67

figure 3.2.10
BaO diffusion in AZS
354 glass, 1425 C
BaO weight ..
12,-----------------,-----------------,

10

8
AZS glaaa
6

2r
o ~_~ ~ __L . - L _ + _ - - ' - - - - - - ' - - - - - - L - - - - - ' - - - J - - - - ,
-10 -8 -6 -4 -2 o 2 4 6 8
mm
260 hr. 2x renewed -+- 260 hr. 2x renewed
--- 260 hr. 2x renewed --<3-- 260 hr. 2x renewed

figure 3.2.11
BaO diffusion in AZS
395 glass, 1350 C
8.0 weight ..
12,-----------

10

4
AZS glaaB

o L- L -_ _L-_---'-----. ---'---..L--.~--~-----'--~

-10 -8 -6 -4 -2 o 2 4 6 8
mm

- 100 hr. -+- 260 hr. 2x renewed --Jif- 600 hr. 6x renewed
 68

BaO diffusion in AZS figure 3.2.12

354 glass, 100 hr.

SaO weight ..
12 r - - - - - - - - - - - - - - - - - . - - - - - - - - - - - - ,

10

8
AZS gl. . .
6

O'----------'-----'------__ -A~>L-------'----'---------'----'----------.J
-10 -8 -6 -4 -2 o 2 4 6 8
mm

-+- mean 1350 C ----- mean 1425 C ~ mean 1500 C

BaO diffusion in AZS figure 3.2.13

1426 C, 260 hr.

9.0 .elQht 'II
12 r - - - - - - - - - - - - - - - - - . - - - - - - - - - - - - ,
AZS glaa.
10

O'--------'---_'-----L.-~'I'_'_--L.------'-----'----------'-----'---------'
-10 -8 -6 -4 -2 o 2 4 6 8
mm

- 354 glass -+- 395 glass - - lithium glass

2x renewed
 69

figure 3.2.14
BaO diffusion in AZS
1350 C, 260 hr.
9.0 weight ...
12 r---=-----------,..-----------------,
AZS gla8.
10

8 t:;======t
6

-8 -6 -4 -2 o 2 4 6 8
mm

--+- 395 glass --+- Lithium glass

2x renewed

figure 3.2.15
SrO diffusion in AZS
354 glass, 1425 C
BrO weight 'It
12 ,-------=-----------~----------

10

8
AZS gla.8
6

o
-10 -8 -6 -4 -2 o 2 4 6 8
mm

260 hr. 2x renewed -+- 260 hr. 2x renewed

---- 260 hr. 2x renewed -B- 280 hr. 2x renewed
 70

The general shape of the concentration profiles of SrO is the same

as the shape of the concentration profiles of BaO.
Table 5 in annex 2 shows the measured values. The reproducibility
is illustrated in figure 3.2.15. An example of time dependency of
the SrO concentration profile is presented in figure 3.2.16.
The penetration depth into the AZS increases with the increase in
the test duration. The maximum SrO concentration in the AZS
remains unchanged (within the inaccuracy of the measurements) with
the increase in the test duration. The effect of the temperature
has been shown by figure 3.2.17. Both SrO concentration and the
penetration depth into the AZS decrease as the temperature
decreases.
The effect of the use of different glass types is also very
obvious and presented in figure 3.2.18. Generally, an increase in
the concentration in the molten glass, yields an increase in the
concentration in the AZS (shown by the difference between 354
glass and 395 glass).
At 1350 0 C (figure 3.2.19) and at a smaller degree at 1425°C
(figure 3.2.18), the SrO concentration of the lithium containing
glass in the glassy phase of the AZS is higher compared to 395
glass with only a slightly higher amount of SrO.
Again, the penetration depth of SrO from the lithium glass is
exceptionally large compared to that of 354 glass and 395 glass.
Only during the test with lithium glass, after 260 hours at
1500 o C, the penetration depth of SrO could be much more than 9 mm
if a refractory thickness above this value had been used (annex 2,
table 5).

3.2.1.3 Summary of the measured concentration profiles in AZS.

For the systems presented here, the penetration depth increases

when the temperature increases and/or the test period is
extended.
 71

figure 3.2.16
SrO diffusion in AZS
354 glass. 1425 C
8rO weight 'It
12 r------------------,,--------------~

10
AZS glan
8

mm

-&- mean 100 hr. -b- mean 260 hr.

figure 3.2.17
SrO diffusion in AZS
395 glass. 260 hr.
SrO •• Ight ..
121---------19==========~~

10

4
AZS glBBa

o L-----'---_----"-_--4--L--A----"--4--..--L_ _L-_---'--_ _ --L_ _

-10 -8 -6 -4 -2 o 2 4 6 8
mm

-&- 1350 C -b- 1425 C --H- 1500 C

2x renewed
 72

SrO diffusion in AZS figure 3.2.18

1425 C. 260 hr.

8rO welIM,.,
12 r - - - - - - - - - - - - - - - - - . , - - - - - - - - - - - - - - - - ,
AZS glaaa
10

0'---------'----4-~-----'-''''___f'---"---------'----L-------'------'-----­
-10 -8 -6 -4 -2 o 2 4 6 8
mm

- 354 glass -+- 395 glass ----- lithium glass

2x renewed

SrO diffusion in AZS figure 3.2.19

1350 C, 260 hr.

8rC weight ..
121---------1---:j==========~-1
AZS glaaa
10

0'-----------'------'----4----'----f-----''----------'-------L----'----~-
-10 -8 -6 -4 -2 o 2 4 6 8
mm

-+- 395 giass ----- lithium glass

2x renewed
 73

The penetration depths of sodium and potassium far exceed those

of barium and strontium.
The maximum concentration of potassium and to a less degree
that of sodium decreases when the temperature rises (in the
range from 1350 to 1500°C) and increases when the test periods
are extended.
The maximum concentrations of barium and strontium increase
when the temperature increases, but they remain unchanged
(after 100 hours) when the test periods are extended.
The penetration depth of Na, K, Ba, Sr from the lithium glass
into the AZS, invariably far exceeds those from the 354 glass
and 395 glass.
The concentrations of Baa and SrO in the glassy phase of the
AZS, in the experiments with lithium containing glass, are
higher compared to lithium free glass with the same Baa and SrO
concentration. Since the concentrations of NazO and KzO in the
lithium glass were far lower than in the used lithium free
glasses, it is difficult to make the similar statement for
sodium or potassium only on the basis of the sodium and
potassium concentration profiles measured here.
Later in this chapter and in chapter 4 it will be shown that a
similar concentration increase of sodium and potassium in the
glassy phase of the AZS under the influence of lithium in the
glass melt is very likely.
The concentration profiles of the measured oxides, due to
cation diffusion, do not follow the direct route from the
concentrations of the glass composition to the composition of
the glassy phase of the refractory (occurrence of up-hill
diffusion).

3.2.1.4 Discussion of measured concentration profiles in the AZS.

One potential explanation for the difference in the behaviour of

the lithium glass, with respect to penetration depths and
diffusion rates, is the presence of a large amount of lithium
oxide. The lithium ion diffuses in the AZS, which results in a
 74

strong reduction in viscosity of the glassy phase of the AZS and a

consequent increase in cation mobility. Although no data are
available on the influence of lithium on the viscosity for the
chemical composition range of the glassy phase of AZS, it is
possible to give an estimate by making use of data for TV screen
glass. For instance, the influence on the viscosity per weight %
of Li 20 is approximately 3 respectively 6 times larger than that
of Ha 20 and K20.
An other interesting phenomenon is the presence of up-hill
diffusion in the measured concentration profiles for all four
presented cations (sodium, potassium, barium and strontium). In
the case of potassium oxide, even two peak maximums in one
concentration profile occur (figure 3.2.8). Up-hill diffusion
occurs because in fact, not the gradient in the concentration
profile but the gradient in the chemical potential respectively
the activities of the diffusing cations determine the direction of
the diffusion.
If the chemical potential of the oxides of sodium, potassium,
barium and strontium is lower in the AZS glassy phase, for
instance because of the high AI 20) content compared to the glass
melt, the system will try to shift into the direction of the
lowest Gibbs energy. The cations will therefore diffuse into the
glassy phase of the refractory until at last the activities are
equal.
If the chemical composition of the glass, respectively the glassy
phase of the fused cast AZS in the transition layer is studied,
one sees that the chemical composition does not follow the direct
route from the glass composition to the composition of the glassy
phase of the original fused cast AZS. Dietzel (Lit. 1) explains
this behaviour, by assuming that the glassy phase tries to adopt a
structure that resembles that of a (thermodynamic relatively
stable) crystal in the neighbourhood of the concentration route.
The crystal structure distinguishes itself by its large
thermodynamic stability and low Gibbs energy. Examples are:
crystal formation of sodium nepheline (Ha20.AI20]" 2Si02), leucite
(K20.AI 20 3.4Si02) or kaliophelite/kalsilite (K20.Alp3.2Si02).
 75

A indication of the thermodynamic stability of these crystals is

their high melting point. The melting point of nepheline is 1526
°C, of leucite is 1686 °c and of kaliophelite/kalsilite is 1750 °c.
Figure 3.2.20 shows a maximum in the K20 content in the glassy
phase, at the interface glass melt/AZS, which corresponds with
about 25 weight% AI 20]" The AI 20] weight% of leucite is 23.4%. The
second maximum concentration of K20 deeper in the AZS meets a AI 20]
content of 32 weight%. The Al 20 3 weight% of kaliophelite/kalsilite
is 32.2 weight%. The "unusual" behaviour of the K20 concentration
profile at the interface and deeper in the AZS might therefore be
explained by attempts to adopt the concentration of the
thermodynamically stable leucite and kaliophelite/kalsilite
compositions.
The position of the flat maximum of the Na20 concentration and the
concentration AI 20] of 35 to 38 weight%, agrees very well with the
concentration of AI 20] in nepheline (35.9 weight%).

figure 3.2.20
Concentration profiles
395 glass, 1425 C, 260 hr.
K20. Na20 .eJght .. AI20S "'aM ..
14,-----------------,------- 40

12 35

10 30

8 26

6 20

4 AZS glaso 15

2 10

o L I_ - - - - - ' - _ - - - - - - - ' -_ _ - - '_ '--_L',,=='===='===='---_----.J

_ 6
-10 -8 -6 -4 -2 0 2 4 6 8
mm

- AI203 -+- K20 - - Na20

 78

crucible has been renewed several times during one test.

The reason for the frequent renewal of the molten glass has been,
to maintain the glass composition close to the original glass melt
composition. In this way, the driving force for the diffusion of
the cations in the AZS, the difference of the chemical potential
of the cations between the molten glass and the AZS has been close
to a practical situation.
The glass has been renewed by pouring glass out of the crucible
and refilling it by freshly molten glass. The glass that has been
poured out of the fused cast AZS crucible will be referred to as
renewed glass. This glass has been cooled down to room temperature
and has been analyzed by a method (XRF) described in section
3.2.2.2.
In the experiments of 260 hours duration, the refreshments of
glass melts take place after about 90 and 160 hours. In the
experiments of 600 hours, the refreshments of the glass melt take
place after about 90, 163, 258, 331, 426 and 499 hours.
The chemical composition of the glass has been measured before and
after it has been in contact with the fused cast AZS of the
crucible. The difference in chemical composition can be caused by
evaporation, corrosion products of the crucible refractory and the
diffusion of cations of the glass melt into the refractory
crucible walls.
In the next section it will be shown that evaporation is
neglectable and it is possible to calculate a kind of diffusion
rate of the cations which can be compared for different cases.

3.2.2.2 Measurement method.

Measurements of the renewed glass in the experiments of 260 and

600 hours, have been made with X-ray Fluorescence (Philips PW
1400) using production standard calibration lines, which ensures a
higher accuracy and reproducibility than the SEM/EDX measurements
of part 3. 2 • 1-
Two methods of sample preparation have been used. In the first
 79

method, 5 grams of cullet of the renewed glass, which is poured

out of the crucible during the test, is molten to a bead on a
small platinum disk. The melting time has been 2 minutes at 1000
°c.
A second method is used to increase the reproducibility of these
measurements. In this method 15 grams of cullet of the renewed
glass, is ground to an average grainsize of 5 micrometer. 0.5 gram
of this material is mixed with 2 grams of Li 2B40 7 • The mixture of
glass and Li 2 B4 0 7 has been molten for 5 minutes at 1000 °C.
The Li 20 in the glass cannot be measured with X-ray Fluorescence.
Therefore the Li 20 has been measured with Flame Atomic Absorption
Spectroscopy (Philips SP 9).
For this analysis, 200 mg of the renewed glass is dissolved in HF
and HCl04 and evaporated. The residue is dissolved in 20 ml HN03
(6N) and filled up to 200 ml with water. The lithium in the
resulting solution is measured using a calibration line (Com. Bur.
of Reference).
All measurements have been performed in duplicate.
The difference in chemical composition of the glass before and
after the test has been determined by this way. The difference may
be due to evaporation or diffusion to or from the fused cast AZS,
during residence of the glass melt in the crucible.

Table 3.6

Evaporation of Iilbium glass

lemp. time U20 Na20 K20 BaO SrO A1203 Si02 Cao noz Sb203 PeZ03
C brL _igbl 'JII

0 0 3,8' 3,31 3,73 8,32 8,33 4,96 6'.6 0,96 D,2O 0,38 0.069
3,38 3.76 8,30 8,30 4.93 6'.6 0,96 D,2O 0,38 0.070

13'0 92 3,83 3,24 3,70 8,3' 8,24 4,97 6',9 0,96 D,2O 0,38 0.070
3,3' 3,71 8,34 8,20 4,96 6',8 0.96 0,20 0,38 0.070

1423 94 3,83 3,28 3,70 8,3' 8,23 4,94 6',9 0.96 D,2O 0,37 D.071
3,23 3,72 8,3' 8,22 4.96 6',9 0.96 D,2O 0,37 0.069

1'00 93 3,83 3,34 3,71 8,34 8,26 4,98 66,0 0,96 D,2O 0,3' 0.069
3,23 3,69 8,3' 8,24 4,97 66,2 0,96 D,2O 0,34 0,070
 80

It can easily be shown that the amounts of material that have been
evaporated are within the measuring accuracy for all three glass
types used (see table 3.6 for lithium glass as an example).
So the change in the measured oxide contents in the glass, is
predominantly due to the diffusion of cations and the increase of
the corrosion products of fused cast AZS, like AI 20] and Zr02' The
decrease of all the concentrations of the oxides, except for AI 20]
and Zr02, by the dilution with the corrosion products from the
fused cast AZS is on a relative scale the same for all studied
oxides, except for a small deviation for Na20 caused by the Na20
content in the original AZS (see table 1.1). The amount of renewed
glass in the crucibles and the dimensions and weight of the
crucibles themselves are kept as constant as possible. In this way
the decrease in the concentration of an oxide in the glass is a
relative measure of the total diffusion in the crucible test set-
up. This relative measure of the rate in the diffusion per oxide
is expressed in electron equivalents per hour (e.e.h.). In other
words: the e.e.h. is the decrease in molar content of a certain
oxide of the glass divided by the residence time (in hours) of the
glass during the test and mUltiplied by its valence.
The e.e.h. is a kind of "overall" value of the diffusion rate. The
e.e.h. is a combination of a lot of effects like:
The difference in partial Gibbs energy of the cation in the
melt and the glassy phase of the AZS (driving force).
The mobility of the cation.
The diffusion of AZS material in the glass melt.
The duration of the experiment with the increasing
concentration of cations of the glass melt in the AZS.
The interaction time of the glass melt between glass melt
renewals with the AZS, which causes a decrease of diffusing
cations in the glass melt and will decrease the driving force
of the diffusion of the cations into the AZS.
The wall thickness of the crucible (constant).
The contact area of glass melt and AZS (constant).
The interaction and inter-diffusion of the various cations.
The temperature.
 81

The chemical composition of the glass melt.

The e.e.h. which is determined in this study is a relative measure

for the diffusion rate of a cation in the AZS, and may only be
compared when an identical test set-up and procedure (including
time) is applied. with the standard test set-up and procedure, an
approximation is made of the influence of glass melt composition
and temperature on the diffusion rate of cations (e.e.h.) into new
AZS, with an "infinite" quantity of glass melt and relatively
large wall thickness of the fused cast AZS.
The relative diffusion rates that will be determined are not
consistent with the diffusion theories with a timeo. s dependence
for a number of reasons, for instance the glassy phase of the
fused cast AZS increases during the experiment, the effective
diffusion coefficients are dependent on the chemical composition
of the glassy phase of the fused cast AZS, also the equilibrium
concentrations of oxides of the diffusing cations in the fused
cast AZS are influenced by the diffusion of the cations and
therefore changes during the test time.
In 3.2.2.3 the used assumptions are given to determine the e.e.h.
of lithium, sodium, potassium, barium and strontium for new AZS,
with a large wall thickness and an "infinite" quantity of glass
melt (simulated by refreshing the melt after certain time
intervals).
Annex 3 gives the measured values of Li+, Na+, K+, Ba 2 + and Sr2 + in
electron equivalents per hour, in table form and gives also an
example of the calculation. In the figures of this chapter the
calculated e.e.h. of the oxides of a renewed glass are placed in
the time scale at the average of their residence time in the total
test time, for example when a glass melt is poured in the crucible
at 90 hours after the start of the test and poured out of the
crucible 70 hours later at the total test time of 160 hours, the
measured e.e.h. are placed in the time scale of the figures at [90
(start time) + 70/2 (average residence time)] = 125 hours.
 82

3.2.2.3 Results of glass analysis after refractory interaction

and determination of e.e.h. values.

Figure 3.2.21 shows the determined diffusion rates expressed in

e.e.h. values of Na+, K+, Ba2+ and Sr2+ for a 395 glass melt at
1500 o C. The concentrations of Na20 and K20 show a marked reduction
on the time scale, owing to the fact that the penetration depth
tends to exceed the crucible wall thickness. This means that if
the wall thickness is infinite, the diffusion rate would probably
remain as it was in the first 200 hours.
Relative diffusion rates from 395 glass into fresh fused cast AZS
at 1500 0 C in electron equivalents per hour are:
Na+: 34 e.e.h.
K+ : 48
BaH: 9
SrH : 15

Figure 3.2.22 shows the relative diffusion rates of Na+, K+, Ba2+
and Sr2+ for a 395 glass melt at 1425°C. The diffusion rate of K20
shows a strong reduction on the time scale, owing to the
penetration depth tending to exceed the crucible wall thickness.
The diffusion rate of Na20, however, shows an increase followed by
stabilization, although its penetration depth also tends to exceed
the crucible wall thickness. (The same phenomenon manifests itself
with 395 glass-glass at 1350 o C, figure 3.2.23). The glassy phase
of the AZS contains about 3.5% Na 20 in its original state (local
SEM measurements of untreated AZS 32 (ER 1681». Figure 3.2.3
shows that part of the Na 20 in the AZS diffuses to the interface
because the concentration of Na 20 at the outer wall side is lower
than of the original AZS. After some time the diffusion of Na+
from the glass melt is decisive. In practice the Na20 from the AZS
is a finite source and that from the glass by continuously
charging is an 'infinite' source. Therefore, in this study, the
diffusion rate is derived from the right hand section of the graph
to minimize these initial counter diffusion effects.
 83

Figure 3.2.21 The relative diffusion rates of Na+, K+, Ba 2 + and

Sr2 + for a 395 glass melt at 1500 °c.

Diffusion of cations into AZS Diffusion of cations into AZS

395 glass. 1500 C 395 glass. 1500 C

electron equivalenta per hour by Na+ electron equivalenta per hour by K+

60,------'-------'--------'------------, 60,--'-------'-------'--------'-------,

50 50
"
" "
40 40 " l!
+
+ f
+ +
30
+
30
"" "
20
*+ +
+ 20
" .. "
~

+ ~
"
~
10 10

0 0

-10 -10
0 100 200 300 400 500 0 100 200 300 400 600
hour.. hour..

~ NII20 " K20

Diffusion of cations into AZS Diffusion of cations into AZS

395 glass. 1500 C 395 glass. 1500 C

.lectron equiv.tenta per hour by 8a2+ electron equiva.ent. per hour by Sr2+
60,----'-----'-----'-----'-----'-----'-----'-----'----c:.:...::..::.::--------, 60 r::..:::..::.::....:c=:::..:::..:-'------::.::-'----::..:::..:-=..:...-'--'-----,

50 50

40 40

30 30

20 20

10 8 iii II B 0 10
0 ~

0 0

-10 -10
0 100 200 300 400 500 0 100 200 300 400 500
hours. hour..

0 BaO x SrO
 84

Figure 3.2.22 The relative diffusion rates of Na+, K+, Ba 2+ and

Sr2 + for a 395 glass melt at 1425 °c.

Diffusion of cations into AZS Diffusion of cations into AZS

395 glass. 1425 C 395 glass. 1425 C

electron equivalents per hour by Na+ electron equivalents per hour by K+

60,------'-------'-----'------------, 60

60 60

40 40

""
30 30
" "
20 20 '"
+ + " Ii t
+
10 10
+
:t: +
0f--------------------1 o
-10 '-----_ _-'--- -'--_ _---i '---_ _
-10
o 100 200 300 400 600 o 100 200 300 400 600
hours. hour•.

+ Na20 " K20

Diffusion of cations into AZS Diffusion of cations into AZS

395 glass. 1425 C 395 glass. 1425 C

."etro" equivalents per hour by 882+

60,------'-------'-----'-----------,
60 relec:.c.:.:'.:.:,o:.:.o:.:.e:..:q:..:U:.:.iv=elc:.eo:.:.'c:.a-.::pc:.e':.:.h:.:.o:..:u:.:.,cbY
:.'---'-Sr_2_+ ----,

60

40 40~-------------------------- ------------------------- ---------------------- -- --------------- I

30 30

20 20
x
x x
10 0
10 x x
x
8 0 CI c B x
B
0 0f---------------------1

-10 '-----_ _--'-_ _---'-_ _---i -'----_-----'

-10
0 100 200 300 400 600 o 100 200 300 400 600
houri. hour..

0 BaO x S,O
 85

Figure 3.2.23 The relative diffusion rates of Na+, K+, Ba 2+ and

Sr2+ for a 395 glass melt at 1350 °e.

Diffusion of cations into AZS Diffusion of cations into AZS

395 glass. 1350 C 395 glass. 1350 C

eo electron equivalents per hour by He. electron equivalent. per hour by K+

eo ,------'-------'----'------------,

50

40 40
"
30 30 "
"" I
20 20 " "
" "
10
+
~
~
~

~
10
"" •"
± + ~
0 0
+
,., +
t
-10
+ -10
0 100 200 300 400 500 0 100 200 300 400 500
hours. hours.

~ Na20 " K20

Diffusion of cations into AZS Diffusion of cations into AZS

395 glass. 1350 C 395 glass. 1350 C

80 ••elm" equivatenta per hour by B_2+ electron equivalent. per hour by 81'2+
eo ,-----"-----'-----=---------,

50 50

40 40

30 30

20 20

10 10 x x
II x
(]
(] (]
~ x
(] ~
II n
(]
§ il
0 0

-10 -10
0 100 200 300 400 500 0 100 200 300 400 500
hour•. hour•.

(]
BaO x S,O
 86

The penetration depth of sodium tends to exceed the crucible wall

thickness (see figure 3.2.5) at this temperature after about 260
hours, therefor the in this way determined relative diffusion rate
for sodium is probably somewhat too low.
The relative diffusion rates of 395 glass components into fresh
fused cast AZS, at 1425°C in electron equivalents per hour are:
Na+: 14 e.e.h.
K+ 34
Ba2+: 5
2
Sr +: 10

Figure 3.2.23 shows the relative diffusion rates of Na+, K+, Ba2 +
and Sr2 + for a 395 glass melt at 1350°C.
The behaviour of Na 20 and K2D resembles that at 1425°C, the only
difference is the even more exceptional behaviour of Na2D. At
first Na ions even diffuse from the AZS into the glass. The BaO
and SrO analysis show large variations, even after the samples had

BaO diffusion in AZS figure 3.2.24

395 glass, 260 hr.

8.0 weight lit
12.-----------------,--------------,

10

4
AZS gl•••

O'------......L---i*'--~-__II_L------'----L------'----.L--­
-10 -8 -6 -4 -2 o 2 4 6 8
mm

--+- 1350 C -4>r- 1425 C - - 1500 C

2x renewed
 87

been ground and mixed. This may indicate that there is a strong
degree of segregation. The amount of barium diffusing into the AZS
at 1350 0 C is much smaller than the amount diffusing at 1425 and
1500 0 C (see for example figure 3.2.24).
The area under the curves in the graphs in this figure is a factor
of two to three times smaller at 1350 0 C than at 1425°C. The
measurements carried out after the longest test period (figure
3.2.23) yield the smallest variation, the diffusion rate of Ba2 +
at 1350 °c is nearly a factor of two lower than the diffusion
measured at 1425°C.
The amount of strontium diffusing into the AZS at 1350 0 C is far
lower than the diffusion rate of strontium at the higher
temperatures, 1425 and 1500 0 C (see for example figure 3.2.17). The
area under the curves in the graphs in this figure is three to
seven times smaller at 1350°C than at 1425°C.
The measurements carried out after the longest test period (figure
3.2.23) yield the smallest variation, the diffusion rate is nearly
a factor of two and a half lower than the diffusion rate at
1425°C. Diffusion rates of components of 395 glass into the fresh
fused cast AZS, at 1350 0 C in electron equivalents per hour are:
Na+: 4 e.e.h.
K+ : 28
Ba2+ : 2
Sr2+ : 3

Figure 3.2.25 shows the diffusion rates of Na+, K+, Ba2 + and Sr2 +
for a 354 glass melt at 1425°C. The diffusion rate of K20 shows a
marked reduction on the time scale, owing to the penetration depth
tending to exceed the crucible wall thickness.
The diffusion rate of Na20 shows a light tendency to drop.
Those of BaO and SrO, again, show reasonable stability on the time
scale.
 88

Figure 3.2.25 The relative diffusion rates of Na+, K+, Ba 2 + and

Sr2 + for a 354 glass melt at 1425 °c.

Diffusion of cations into AZS Diffusion of cations into AZS

354 glass. 1425 C 354 glass, 1425 C

.lectron equivalent. per hour by Na+ electron equivalent. per hour by K+

60,------------'----'-----------, 60r..-------'-------'-------'------,

60 60

40 40

II': .
30 30

+ + 1" I
t + t t +
20
I •
... 10
•
0

-10 -10
0 100 200 300 400 600 0 100 200 300 400 600
houra. hours.

+ Na20
" K20

Diffusion of cations into AZS Diffusion of cations into AZS

354 glass. 1425 C 354 glass. 1425 C

electron equivalent. per hour by 882. 60 8r:1=-8c:..:t::..:ro::..:n::..:8:..::Q=.ui::..:V&:..:1:..:8n::..:I=-&-,=-p:..:8r::..:h::..:o::..:u::..:r::..:bY

:..::..::.Sr::..:2::..:+ ---,
60,------'-------'----'-----------,

60 60

40

30 30

20

10
o o o

_10L---~--___'__-_----' '---_-----.J
200 300 400 600 o 100 200 300 400 600
hourL houra.

o Baa x 8rO
 89

Diffusion rates of 354 glass at 1425°C into fresh fused cast AZS
in electron equivalents per hour are:
Na+: 20 e.e.h.
K+ 26
Ba2+: 6
sr 2 +: 3
Figure 3.2.26 shows the relative diffusion rates of Na+ and K+ for
a 354 glass melt at 1350 0 C.
Since there are only two experiments done, there are only two
measured e.e.h. values for each element, the reliability of the
resulting e.e.h. is lower than usual.
The approximate relative diffusion rates for the case of 354 glass
into fresh fused cast AZS at 1350°C given in electron equivalents
per hour are here determined:
e.e.h.

Figure 3.2.26 The relative diffusion rates of Na+ and K+ for a

354 glass melt at 1350 0 C.

Diffusion of cations into AZS Diffusion of cations into AZS

354 glass, 1350 C 354 glass, 1350 C

80;::.Ie=c:..:.'ro:=n-=-.q~u::..:iv-=-al-=-.n::..:'-=-.-=-P.:..:.'::..:h::..:ou:..:.'-=-bY'--.:..:.N.-=-. ~ 80 ;:.I.....C:..:.":-.o:..:.n-=-.q~u:::iv-=-al=.n::..:'a-=--=-p.:..:.'::..:h.::ou:..:.'-=-by'--.:..:.K. _

60

40 40

30 30

20 "
"
10

01------------------1

_ 10 I---_----l'---_----'_ _----'_ _----'-_ _-.J

200 300 400 600 o 100 200 300 400 500
hour•. hourL

+ N.20 " K20

 90

Figure 3.2.27 The relative diffusion rates of Na+, K+, Ba 2 + and

Sr 2+ for lithium containing glass at 1500 °C.

Diffusion of cations into AZS Diffusion of cations into AZS

Lithium glass. 1500 C Lithium glass. 1500 C

electron equivalents per hour by Na+ electron equivalents per hour by K+

60r---~-- 60,------'-----'------''------------,

60 60

40 40

30 30

20 20 " "
+
10 10

0f---------------------1 0

-10 ~--'-----'---~ -'0

o 100 200 300 400 500 0 '00 200 300 400 600
houre. hour•.

+ Na20 " K20

Diffusion of cations into AZS Diffusion of cations into AZS

Lithium glass. 1500 C Lithium glass. 1500 C

electron equivalents per hour by B82. electron equivalents per hour by 8r2·
60,-----------'-------------, 60 r-:-=-=--'--=-cc=-'-=--'-'------------,

60 60

.
40 40

.
30 30

20 o o 20

10 10

o f---------------------j

-'0 -----'-----'-------'---~'----
o 100 200 300 400 600
hours, hour•.

U BaO x BrO
 91

Figure 3.2.27 shows the relative diffusion rates of Na+, K+, Ba2 +
and Sr 2 + for lithium glass at 1500 0 C.
since, again, there are only two experiments done, only two
measured values for each element are available, the reliability of
the resulting e.e.h. is lower than for the 395 glass.
Na20 shows an increase in the diffusion rate values on the time
scale at all the three temperatures (figures 3.2.27, 3.2.29 and
3.2.30) for this case.
This effect is probably caused by the low Na 20-content of the
glass (annex 1). Due to this low Na 20-content of the glass at the
start of the experiments the Na+ will diffuse from the AZS to the
glass melt interface. This is the reason for the lower average Na+
diffusion from the glass melt into the AZS during the first 90
hours. This is why, for lithium containing glass, the Na+
diffusion rate, determined after a melting time of 125 hours is
the best estimate.
The penetration depth of K+ in lithium containing glass is so
large, that the diffusion rates measured in this way are actually
lower than the values when the penetration depth would not exceed
the wall thickness.
The relative diffusion rate of SrO shows a drop on the time scale
(figure 3.2.27). Figure 3.2.28 shows that the penetration dept
after 260 hours at 1500 °C just tends to exceed the crucible wall
thickness. since the glass used for the second measurement has
been poured into the crucible after 160 hours, the effect of a
finite wall thickness will have had only a slight effect. The
concentration of Li 20 (figure 3.2.31) cannot be measured by using
the SEM/EDX. If, however, the penetration depth of Li+ is assumed
to be the same as that of Na+ or K+ because of its small size, the
shortest period would probably give the best indication. The
measured relative diffusion rates into fresh fused cast AZS of
lithium containing glass components, at 1500 0 C in electron
 92

SrO diffusion in AZS figure 3.2.28

Lithium glass, 260 hr.

8rO welliJht t.
12,-------------------,--------------,
AZS glass
10

OL-----'------6--'------4!----'--------'----'-------.L_ _--L-_----!
-10 -8 -6 -4 -2 o 2 4 6 8
mm

--+- 1350 C --A-- 1425 C ~ 1500 C

2x renewed

equivalents per hour are:

Li + : 86 e.e.h.
Na+ : 22
K+ 22
Ba2+: 19
Sr 2 +: 28

Figure 3.2.29 shows the diffusion rates of Na+, K+, Ba 2 + and Sr2 +
for a lithium glass melt at 1425°C.
The pattern is identical with that of figure 3.2.27 at 1500 o C,
with the difference that the penetration depth of SrO is smaller
than the crucible wall thickness (figure 3.2.28).
Figure 3.2.31, in turn, shows a drop in Li 20-diffusion on the time
scale, at a temperature of 1425 DC. This indicates that the
 93

Figure 3.2.29 The relative diffusion rates of Na+, K+, Ba2 + and
Sr 2 + for lithium containing glass at 1425 °e.

Diffusion of cations into AZS Diffusion of cations into AZS

Lithium glass. 1425 C Lithium glass. 1425 C

electron equivalents per hour by Ha. electron equivalents per hour by K+

80 r - - - - - - ' - - - - - - - - ' - - - - - - - - ' - - - - - - - - - - , 80 r - - - - - - ' - - - - - - - - ' - - - - - - - ' - - - - - - - - - - ,

60 - 60

40 40

30 30

20 20

10 + 10

0
+ of-----------------------1

-10 _10L----'-------'-------'----L-------.J
0 100 200 300 400 600 o 100 200 300 400 600
hour•. hour•.

+ Na20 .. K20

Diffusion of cations into AZS Diffusion of cations into AZS

Lithium glass. 1425 C Lithium glass. 1425 C

bY
80 ,.,_.c.:...'.:...'o.:..."_.:..:Q"'u.:...iv-=.al.:....".:...'.:...8-.::P.:....'.:...-=-hO:..:u.:...'.:...:.....:cB-=.82.:...+ -----, electron equivalent. per hour by 81"2.
80 r=-.:....:....:....:...:..::::...:..::.:.:....:....:...-.::.:....:....:...:..:.:....:...:.....:c.:....:...-------,

60 60

40 40

30

.......
20 20

10 o 10

of-----------------1 of-----------------------1

_10L-_ _--'---_ _---'--_ _---' -'----_-----.J -10L----'-------'--------'-------'-------'

o 100 200 300 400 600 o 100 200 300 400 600
hour•. hourL

o Baa x 1r0
 94

Figure 3.2.30 The relative diffusion rates of Na+, K+, Ba 2 + and

Sr 2 + for lithium containing glass at 1350 °c.

Diffusion of cations into AZS Diffusion of cations into AZS

Lithium glass. 1350 C Lithium glass. 1350 C

eo electron equivalent. per hour by N.+ electron equivalent. per hour by K.

80,------'---------------,

50 50

40 40

30 30

20 20

10
" "
10

01-----------------1
_ 101-----'---~-----'---L--­
o 100 200 300 400 500 o 100 200 300 400 500
hours. hOUf••

+ Na20 " K20

Diffusion of cations into AZS Diffusion of cations into AZS

Lithium glass. 1350 C Lithium glass. 1350 C

e.ectron equivalent. per hour by S&2+ electron equivalent. per hour by 8,2.
80,-----'-----'----'----------, 80,------'-------'----------,

.......
50 50

40 40

30 30 .

20 20

10
10r=0

o
- 10
- -----------j----j
----'----------~
of-----------------------1
_ 10 L I_ _- - ' - -.L-_ _- - ' -_ _- - ' -_ _

o 100 200 300 400 500 o 100 200 300 400 500
houra. hour•.

o BaO x 8rO
 95

Figure 3.2.31 The relative diffusion rates of Li+ for lithium

containing glass at 1350, 1425 and 1500 ·C.

Diffusion of cations into AZS Diffusion of cations into AZS

Lithium glass. 1500 C lithium glass. 1425 C

100 ~el=-ec::l::.:ro:::n-=e~qu::i.::va=le:::n:::I.~p::e::.-r
::.:ho::u=_r-=b~Y.:.li:-+ -, 100 ~el:::ec::l:.:ro:::n-=e~qu:::lv:.:a:::le:::n::.:I.~p:.:e::.-r
:::ho::u::.-r-=by~Li:-+ -,

90 90
80 80
70 70 "
80 80
60 60
40 40

30 30
20 20
10 10
o~----------------..j 0~-------------___1
-10 L -'-_ _--'-_ _--'--_ _---'---_ _...J -10 "----'------'-----'-------'--------'
o 100 200 300 400 600 o 100 200 300 400 600
hour•. hour••

x Li20 x Li20

Diffusion of cations into AZS

Lithium glass. 1350 C

100 ~el=e::cl::.:ro:::n::.:e:..:q::ui:::va:::le:::n:::I.:..:p::.=-r
:::h::ou::.r_=b!...y.:.L:..-i+ _

90

80'
70

80

60
40

30

20

10
ol--------------~

-10 L-_ _.L-_ _-'--_ _- ' -_ _--'-_----l

o 100 200 300 400 600
hour•.

x Li20
 96

crucible wall is probably too thin.

The relative diffusion rates determined in this way for the case
of lithium containing glass into fresh fused cast AZS at 1425°C in
electron equivalents per hour are:
Li + : 58 e.e.h.
Na+: 10
K+ 19
Ba2+ : 11
Sr 2 +: 17

Figure 3.2.30 shows the diffusion rates of Na+, K+, Ba 2 + and Sr2 +
for lithium glass at 1350°C.
The patterns are similar as those in figures 3.2.27 and 3.2.29.
Figure 3.2.31, again, shows a drop in Li 20-diffusion on the time
scale, which is very slight at 1350 0 C.
The relative diffusion rates for the case of lithium containing
glass into fresh fused cast AZS at 1350°C in electron equivalents
per hour are:
Li + : 34 e.e.h.
Na+ : 3
K+ : 14
Ba2+ : 7
Sr 2 +: 11

3.2.2.4 Discussion.

Table 3.8 summarizes the measured relative diffusion rates of

cations into fresh fused cast AZS in electron equivalents per hour
for the different cases presented in this study.
This table clearly shows that the relative drops in potassium
diffusion rates by decreasing temperatures, are smaller than the
relative drops in the diffusion rates for the other elements. The
amount of diffused cations in the AZS is a combination of the
fraction of glassy phase in the fused cast AZS, the concentration
of the cation in the glassy phase and the penetration depth.
 97

Table 3.8

cation Li+

glass type 354 glass 395 glass Li

1350 0 e * * 34

1425°e * * 58

1500 0 e * 86
*
wt% in original glass as Li 20 0 0 3.84

mol% in original glass as L~O 0 0 8.43

cation Na+

glass type 354 glass 395 glass Li

1350 0 e 4 4 3

1425°e 20 14 10

1500 0 e 34 22
*
wt% in original glass as Na 20 9.43 6.84 3.30

mol% in original glass as Na 20 10.34 7.72 3.50

cation K+

glass type 354 glass 395 glass Li

1350 0 e 18 28 14

1425°e 26 34 19

1500 0 e * 48 22

wt% in original glass as K20 6.46 9.10 3.72

mol% in original glass as K20 4.66 6.75 2.59

 98

cation Ba2+

glass type 354 glass 395 glass Li

1350 0 C 2 7
*
1425°C 6 5 11

1500 0 C 9 19
*
wt% in original glass as BaO 10.75 7.70 8.34

mol% in original glass as BaO 4.76 3.51 3.57

cation Sr2+

glass type 354 glass 395 glass Li

1350 0 C 3 11
*
1425°C 3 10 17

1500 0 C 15 28
*
wt% in original glass as SrO 3.01 9.59 8.25

mol% in original glass as SrO 1. 97 6.47 5.23

The difference in behaviour of the potassium can be explained from

the difference in penetration depths and the maximum potassium
concentrations which can be absorbed in the AZS.
The penetration depths of all the cations drop as the temperature
decreases. The relative penetration depth of sodium, however,
drops stronger than those of the other cations, such as potassium.
See figure 3.2.32, which shows the penetration depths of the
maximum concentrations after 100 hours at different temperatures.
The maximum concentrations of BaO and SrO in the glassy phase of
the AZS drop strongly as the temperature decreases, while those of
Na 20 and K20 rise, although the maximum concentration of K20
increases at lower temperature much stronger than that of Na20
(see figure 3.2.33).
 99

The two effects, decrease in penetration depth and maximum

concentration bring about a drop in the "diffusion rates" of BaO
and SrO determined in this way when the temperature decreases.
The reduction in penetration depths of Na 20 and K20 gives on one
hand a reduction effect in the measured diffusion rates, and the
increase in the maximum concentrations in the AZS on the other
hand, would give a rise in the diffusion rates as the temperature
decreases. As the rise in maximum concentration of Na 20 (figure
3.2.33) is far smaller than the exceptional relative reduction in
penetration depth (figure 3.2.32), the diffusion rate of sodium
shows a strong reduction as the temperature goes down.
The rise in maximum concentration of K20 (figure 3.2.33), however,
is so high that it offsets part of the reduction in penetration
depth (figure 3.2.32), which implies that the diffusion rate going
with a drop in temperature shows a far smaller reduction than in
the case of the other elements.
The strong rise in the e.e.h. values in the lithium glass is
probably due to the diffusion of lithium in the glassy phase of
the fused cast AZS having a significant influence on the viscosity
of the glassy phase of the AZS and therefore on the diffusion of
the other cations. The first part of this chapter shows that the
penetration depth of lithium glass invariably far exceeds those of
the situations for 354 and 395 glass. Another probable cause is
the increase in equilibrium concentration of the cations in the
glassy phase of the AZS with the cations in the glass melt due to
the Li 20 diffusing into the AZS (see section 3.2.1.3 of this
study), which increases the driving force for diffusion. Also the
increase of the glassy phase of the AZS due to the diffusion of
lithium, which dissolves the corundum (Al 20 3 ) crystals, might be a
cause for the larger e.e.h. values in the case of lithium glass
melt AZS interaction.
The here determined relative diffusion rates are a measure for the
flux of the positive charge transported by the cations in the
glassy phase of the fused cast AZS.
 100

Penetration depth of maximum figure 3.2.32

concentrations peaks in
AZS-354 glass after 100 hrs.
mm.
5
0

0
3 .........

2 0

"
0
1300
"
1350 1400 1450 1500 1550
temperature

0
" Na20 K20

Maximum concentrations in figure 3.2.33

AZS-354 glass after 100 hrs.
weight '10
12,-------------------------------,

10

0"-----"------
1300 1350 1400 1450 1500 1550
temperature

- - Na20 -+- K20 - - BaO --a- SrO

 101

The positive charge transported in the AZS is the rate determining

step of the oxygen forming mechanism at the refractory/glass melt
interface.
As already is mentioned in 3.2.2.2, the determined relative
diffusion rates are not consistent with the diffusion theories
with a timeo. s dependence for a number of reasons, for instance the
glassy phase of the fused cast AZS grows during the experiment,
the effective diffusion coefficients are dependent on the chemical
composition of the glassy phase of the fused cast AZS, also the
equilibrium concentrations of oxides of the diffusing cations in
the fused cast AZS is influenced by the diffusion of the cations
and therefore changes during the test time.
Summarizing it can be concluded that the relative diffusion rates,
into fresh fused cast AZS, of the most important cations, for the
case of T.V. screen glass have been determined in so called
e.e.h., which is the positive charge transported by the cations in
the fused cast AZS. The transported positive charge is
proportional with the formation of oxygen at the interface fused
cast AZS/glass melt and therefore important for the formation of
bubbles and knots.
In chapter 4 the here determined values will be used in a model
which predicts the formation of bubbles at the interface glass
melt/fused cast AZS.
 102

Literature references chapter 3.

[ 1] A. Dietzel;
Zusammenhang zwischen Phasendiagramm, Reaktionsverlauf und
Struktur von Schmelzen.
Glastech. Ber. 40 (1967) p. 378-381.
[ 2] A.R. Cooper;
Model for mUlti-component diffusion.
J. Phys. Chern. Glasses 6 (1965) p. 55-61.
[ 3] A.K. Varschneya, A.R. Cooper;
Diffusion in the system K2o-Sro-sio2 : III, Inter-diffusion
coefficients.
J. Am. Ceram. Soc. 55 (1972) p. 312-318.
[ 4] P.K. Gupta, A.R. Cooper;
The D matrix for mUlti-component diffusion.
Physica 54 (1971) p. 39-59.
[ 5] V.A. Zhabrev, A.I. Isakov;
The inter-diffusion in silicate melts containing three
cations.
Fiz. Khim. Stekla 12 (1986) p. 188-193.
[ 6] V.A. Zhabrev, A.I. Isakov, G.A. Shashkina;
A study of three-cation diffusion with the interaction of
alkali silicate melts.
Fiz. Khim. Stekla 12 (1986) p. 596-601.
[ 7] K. Hunold, R. Brueckner;
Chemische Diffusion van Natrium- und Aluminimionen in
Natrium-Alumosilicatschmelzen.
Glastech. Ber. 53 (1980) p. 207-219.
[ 8] V.A. Zhabrev, A.I. Isakov;
Criteria determining the form of the diffusion path in
glassy melts with three mobile cations.
Fiz. Khim. Stekla 14 (1988) p. 423-428.
[ 9] A.R. Cooper, A.K. Varshneya;
Diffusion in the system K20-SrO-Si0 2 : I, Effective binary
diffusion coefficients.
J. Am. Ceram. Soc. 55 (1972) p. 312-318.
 103

[10] F. Novotny, M. Dragonn;

Anwendung der irreversiblen Thermodynamik auf die Loesung
der festen Phase in der Glasschmelze,
Silikattechnik 40 (1989) p. 373-376.
 104

4. Effect of glass composition and temperature on glass defect

potential originating from refractory.

4.1. Introduction.

According to chapter 2 the diffusion of cations appears to be the

most probable rate determining step in the oxygen bubble forming
mechanism at the refractory/glass melt interface.
In chapter 3, the diffusion process of the most important cations,
for the case of T.V. screen glasses, into the glassy phase of the
AZS was treated. The diffusion rates of the cations are
experimentally estimated in so called electron equivalents per
hour (e.e.h.), which is the positive charge transported by the
cation into the fused cast AZS, determined in a standardised test
set-up.
In the used test set-up all the crucibles were prepared using new
fused cast AZS 32 material (ER 1681) and were, within small
tolerances, of the same size and weight. Also, the amount of glass
was kept as constant as possible. In this way it is not only
possible to compare the diffusion rates of the different cations
during one test, but also the diffusion rates of the cations of
different tests, following the standardised procedure. In the
standardised test procedure, three different glasses have been
used with different chemical compositions and viscosities (annex
1). Of course, no absolute values of diffusion coefficients could
be obtained with this test. with the measured relative values of
the diffusion rates, it is possible to determine the influence of
the chemical composition and viscosity on the formation rate of
bubbles from the fused cast AZS, within the composition range of
the glasses investigated.
The determined relations for the cation diffusion rate are used in
a model which predicts the formation of bubbles at the interface
glass melt/fused cast AZS. The creation of a model, including the
required diffusion rate data, which predicts the glass defect
potential of different T.V. screen glass melts in contact with
fused cast AZS under different conditions, is one of the most
 105

important objectives of this study.

4.2. Results of measured cation diffusion.

- effect of composition, kind of cation, temperature level
and viscosity of the glass melt.
The diffusion rate of a cation into the AZS is determined by its
mobility and concentration in the AZS glassy phase in equilibrium
with the glass melt. The temperature has a direct influence on the
viscosity of the AZS glassy phase. The composition and therefore
the viscosity of the glass melt can have an indirect influence on
the viscosity of the AZS glassy phase as well, by the glass melt
composition dependent diffusion of glass melt cations into the AZS
which changes the composition of the glassy phase of the AZS.
The diffusion rates in electron equivalents per hour, measured in
the standard crucible set-up, can be represented in various ways.
Although the diffusion of cations in the AZS glassy phase has been
quantified, instead of listing the cation names (~+) , the names
of the cation oxides (MO~) will be given.

Figure 4.2.1 The relative diffusion rate for lithium.

Lithium figure 4.2.1.8 Lithium Ilgure 4.2.1.b

electron equivalents per hour by Li+ 12 e,l_ec_tr_o_n_e--,-qu_IV_a_le_nt_o--,-p_er_h_o_u_r--,by_L_I_. -,

12 ,-:-~----,---:..::c..::..::c..:"'::":''-=-''-'--=-------7l

10 10
a

8 8
a
8 8

4 4 a

2 2

O'--------'-----'--------'------'------..J O'-------_----J'-- ----' ---'----..J

1300 1360 1400 1460 1600 1660 o 6 10 16
lemperature (e) dyn. vlaooalty (PLo)

-e-- Lithium gl••• a Lithium gl...

per mol per cent LI20 per mol per CMt LI20
 106

Figure 4.2.2 The relative diffusion rate of sodium, potassium,

barium and strontium in e.e.h. versus temperature.

Sodium figure 4.2.2.8 Potassium figure 4.2.2.b

:----~--. ------'-'--'---l
electron equivalents per hour by NB+
10
electron equivalents per hour by K+
.-"-'''-'=~'-'--------~

1
0
1300 1350 1400 1450 1500 1550 1300 1350 1400 1450 1500 1550
temperature (e) temperature (e)

+ 35401a .. .. 395 glas8 o Lithium glass + 354 glass .. 395 glass --e- lithium glass

per mol per cent Na20 per mol per cent K20

Barium figure 4.2.2.c Strontium 'igure 4.2.2.d

electron equivalents per hour by 8a2+ electron equivalents per hour by 8r2+
8·

11

0' o
1300 1350 1400 1450 1500 1550 1300 1350 1400 1450 1500 1550
temperature (C) temperature Ie)

+ 354 glU, .. 395 glsss [) lithium glus + 354 glau .. 395 glau o lithium glass

per mol per cent BaO per mol per cent SrO
 107

Figure 4.2.3 The relative diffusion rate of sodium, potassium,

barium and strontium in e.e.h. versus viscosity.

Sodium figure 4.2.3.8 Potassium figure 4.2.3.b

electron equivalents per hour by Na+ electron equivalents per hour by K.

7 --------- 10 - ------ --

o
6
6
o

4 " 6
+0

3
"
4 +

"

* o L -_ _-"--_ ___'_ '---_ _- ' -_ _---"

O~---'-----'----'-----'-------l
o 5 10 15 20 25 o 10 15 20 25
dyn. viscosity (Pa.sl dyn. viscosity (Pa.s)

+ 354 gina '* 395 gla" o Llthium glass + 364 gl... • 396 gja .. o Lithium gl.1I

per mol per cent Na20 per mol per cent K20

:r'-' . . .".,
Barium figure 4.2.3.c. Strontium figure 4.2.3.d

.~,.~, ~~"
electron equivalents per hour by 8r2+
6 ;--:-'---'--'---------'---'------'---'----------,

4
41
3t 3
I

:~-------'----_.J
o 10
dyn. viscosity (Pa.s)
15 20
O~----'------'------'------'
o 5 10
dyn. viscosity (Pa.s)
15
"
20

f- 354 gl . . . .. 395 glass n Lithium glass .. 395 gl . . . n Llthium 01 ...

per mol per cent BaO per mol per cenl SrO
 108

This has been done because all measurement results are in oxide
fractions (given in mol% or weight%) .
Figure 4.2.1, 4.2.2 and 4.2.3 show the e.e.h. values of Li 20, Na20,
K20, BaO and SrO in electron equivalents per hour (e.e.h.) per mol
per cent in the glass composition, against the glass temperature
and viscosity. Figure 4.2.1.a shows a linear relationship between
the e.e.h. of Li 20 and the temperature.
There is no linear relationship between the viscosity and the
e.e.h. of Na 20 (figure 4.2.3.a). The e.e.h. values for the cases
of 395 and 354 glass are identical at a specific temperature. The
e.e.h. values for Na, K, Ba and Sr using the lithium containing
glass are markedly higher than those of 395 and 354 glass at a
specific temperature (figure 4.2.2.a). For potassium there is a
linear relationship between the e.e.h. values and temperature
(figure 4.2.2.b). At a given temperature the e.e.h. values for
potassium of 354 and 395 glass are identical and that of the
lithium glass is higher. The e.e.h. values of barium and strontium
show an identical tendency. Even at identical temperatures there
is a wide difference in e.e.h. between 395 and 354 glass on one
hand and lithium glass on the other hand (figure 4.2.2.c and
4.2.2.d).
The e.e.h. values plotted against the viscosity, which takes the
shape of a curve with a rather narrow band (for the different
glass compositions) for sodium and potassium (figure 4.2.3.a and
4.2.3.b), strongly diverges from the shapes for barium and
strontium (figure 4.2.3.c and 4.2.3.d). For the same viscosity,
the Ba 2 + and Sr 2 + diffusion rates from the lithium glass are much
higher than from the other glasses.
The graph with the e.e.h. values of the oxides in zero-lithium
glasses plotted against the temperature (figure 4.2.4.a) shows
that the e.e.h. of K20 is highest, far higher than those of the
other oxides, notably at 1350 o C. In glasses with a high Li 20-
content the e.e.h. values of all the four oxides are higher than
at the same temperature, those of BaO and SrO the most strongest
(figure 4.2.4.b). The e.e.h. of Li 20 itself is high too (Li+ is a
very small mobile cation). At 1350 0 C the e.e.h. of K20 is less
 109

Figure 4.2.4 The relative diffusion rate of lithium, sodium,

potassium, barium and strontium in e.e.h.

354/395 glass tigure 4.2.4.& Lithium glass figure .... 2.4.b

electron equivalents per hour electron equivalents per hour

8 12,-----'------------------,

• 10

6
" 8 "
" "o +
""
8
"
4
, a
x

*~ I>

!j! +
0 oL-------''---------'~----''--------'----'
1300 1350 1400 1450 1500 1550 1300 1360 1400 1450 1500 1650
temperature (C) temperature (C)

+ Na20 a BaO x SrO

o L120 + Na20 • K20 a aaO x SrC
" K20

per mol per cent oxide per mol per cent oxide

dominant compared to the other oxides, than in the case of zero-

lithium glasses.
Summarizing, it can be concluded that the determined relative
diffusion rate for the cations in the Li 20-containing glasses are
higher than in the Li 20-free glasses. Both figures 4.2.4.a and
4.2.4.b show an approximately linear relation between measured
e.e.h. values and temperature between 1350 and 1500 °C.
The temperature dependency of the e.e.h. of the different cations
are given by the following relations, which are the best fits of
the values presented in the figures 4.2.4.a and 4.2.4.b.
 110

e.e.h. values per mol per cent for LizO-free glasses, in fused
cast AZS 32 material:

e.e.h. per mol per cent NazO -33.8450 + 0.025288*T R=0.982

e.e.h. per mol per cent KzO -23.3744 + 0.020228*T R=0.980
e.e.h. per mol per cent BaO -17.4911 + 0.013295*T R=0.984
e.e.h. per mol per cent SrO -16.1571 + 0.012364*T R=0.994

e.e.h. values per mol per cent for LizO-containing glasses, in

fused cast AZS 32 material:

e.e.h. per mol per cent LizO -51. 5618 + 0.041122*T R=0.999
e.e.h. per mol per cent NazO -40.2501 + 0.029830*T +
0.127461* [LizO] R=0.975
e.e.h. per mol per cent KzO -23.5880 + 0.020379*T +
0.192780* [LizO] R=0.990
e.e.h. per mol per cent BaO -23.9845 + 0.017852*T +
0.237246*[Li zO] R=0.975
e.e.h. per mol per cent SrO -22.7870 + 0.017017*T +
0.249884* [LizO] R=0.977

in which T is in degrees Centigrade, LizO in mol per cents and R

is the correlation coefficient.
These equations have been verified in the following concentration
ranges:

LizO from 0 to 8.5 mol%

NazO from 3.5 to 10.5 mol%
K 20 from 2.5 to 6.8 mol%'
BaO from 3.5 to 4.8 mol%'
SrO from 2.0 to 6.5 mol%
 111

For TV-screen glasses, these are the main oxides whose cations
diffuse into the glassy phase of the AZS. In the current TV-screen
glasses, other cations diffusing into this phase are only small in
quantity.
The other cations diffusing from T.V. glass melt into the glassy
phase of fused cast AZS, are Ca 2 + and Mg2 +. In a T.V. screen glass
the concentrations of CaO and MgO are about 1 weight%. By way of
estimation, the concentration profiles of CaO and MgO have been
determined at various temperatures and for various periods. Figure
4.2.5 and 4.2.6 show them compared with those of BaO. Figure 4.2.5
shows the Y-axis scale (in weight%) of BaO and CaO so selected
that their weight percentages in the glass coincide. The same
'standardized' graduation shows that the diffusion rate in weight
percentages is a factor of 2.5 to 3 lower for CaO than for BaO.
Figure 4.2.6 shows the concentration profiles of MgO compared with
that of BaO. That of MgO expressed in weight% is a factor of 3.5
to 4 lower than that of BaO. The mol weights of BaO, CaO and MgO
are 153.34, 56.08 and 40.30. The ratio between them is well in
line with the ratio between the diffusion rates per weight% of the
glass composition. For an estimate of the share of CaO and MgO in
the total e.e.h. of the glass, in T.V. screen glass with its low
CaO and MgO content, it is sufficient for this approach, to add
the weight percentages of CaO and MgO to that of BaO for the
calculation of the total e.e.h.

4.3 Practical application of the cation diffusion rules by a

glass furnace model.

The cation diffusion is given in electron equivalents per hour

(e.e.h.), which is the positive charge transported by the cations
in the fused cast AZS in the standard crucible test set-up. The
amount of positive charge transported in the AZS is a measure for
the amount of oxygen formation at the interface glass melt/AZS.
With the experimentally determined cation diffusion equations it
is possible to make an estimation of the relative amount of
transported positive charge by the diffusion of cations.
 112

Figure 4.2.5 comparison of the concentration profiles of BaO and

CaO in the interface glass melt/fused cast AZS.

354 glass, 1350 C figure 4.2.5.& 354 glass, 1425 C Ilgura 4.2.5.b

weight .. 8aO weight .. CaO weight' BaD weight .. CaO

1 2 , - - " - - - - - - - - , - - - - - = - - - - - , 1.8 12,--=----------,------=--------, 1.8

10 1.5 10 1.5

8
1.2

0.9
8

8
A'. .,- 1.2

0.9

4 0.8 4 0.8

2 0.3 2 0.3

oL-----'--_L----'---------'----'-------.Jo o L-_--'----<!Ji-----'-_ _L--_--'--_---'-_-----" 0

4 ~ ~ 0 2 4 8 -8 -4 -2 o 2 4 6
mm mm

-e- SaO ---- CaO ~BaO -+-CaO

100 hr•. 100 hra.

354 glas8, 1500 C 'igur. 4.2.6.0 354 glass, 1425 C Ilgura 4.2.5.d

weight .. BaO weight .. CaO weight .. BaO weight .. CaO

12 ,-~------_._----~-----, 1.8 12 r-=:....:c~c::....---_._-----=----_, 1.8

10 1.5 10

8
A" 8'."
1.2

0.9
8
... 8 ....
1.2

0.9
8

4 0.8 4 0.8

2 0.3 2 0.3

o L -_ _ --~---L---L-----'---~ 0 oL----a--'--"'-------L--L-----'-------'-------" 0
4 ~ ~ 0 248 -8 -4 -2 o 2 4 6
mm mm

--&- BaO ---Ii- CaO ---e- SaO ---H--- CaO

100 hra. 260 hra.

 113

Figure 4.2.6 Comparison of the concentration profiles of BaO and

MgO in the interface glass melt/fused cast AZS.

354 glass, 1350 C figure 4.2.8.• 354 glass, 1425 C figura 4.2.6b

weight 'It BaO weight 'It MgO weight .. B.O weight" MgO
1 2 , - " - - - - - - - - - , - - - - - = - - - - - - - ' ' - , 1.2 12,--=---"------,-------'-=-----=, 1.2

10 10

s 0.6 8 0.6

6 0.6 6 0.6

4 0.4 4 0.4

2 0.2 2 0.2

0 0 0 0
-6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6
mm mm

--e-- 8aO -"- MgO --e- BaO -"- MgO

100 hr•. 100 hr•.

354 glass, 1500 C figure 4.2.6.c 354 glass, 1425 C figura 4.2.6.d

weight .. BaO weight 'It MgO weight 'llo BaO waight 'llo MgO
12,--=-------,-------"----'---=, 1.2 12,---=----'--"------,-------=-----=, 1.2

10 10

6 0.8 8 0.8

6
A" .,...
0.6 6
.z. .,- 0.6

4 0.4 4 0.4

2 0.2 2 0.2

o L-_-4-~---'---------L----'-------.JL----' 0 OL--<hI<'_ __'_____ ____L_ ___'___ ____.JL-_ __'

0
-6 -4 -2 o 2 6 -6 -4 -2 o 2 4 6
mm mm

--e- BaO -"- MgO --e- BaO ...... MgO

100 hra. 260 hra.

 114

To make such an estimation besides the chemical composition of the

glass melt the contact temperature of the glass melt/fused cast
AZS has to be known.
To illustrate the application of the model, let us assume a
typical T.V. screen glass furnace with a glass output of 1.2 to
1.4 tons per day per mm2 melting end surface, the industrial
furnace is 19.4 m long, 6.7 m wide and its glass level 1.02 m
high. The contact temperatures that will be used in the example
have been checked with physical temperature measurements combined
with computer model simulations.

Figure 4.2.7 In this picture of the glass bath of a melting

end, the numbers give the contact temperatures
in degrees Centigrade in the example of the
model.
 115

The temperature of the whole bottom is assumed to be 1350 o C, that

of the metal line (the three phase contact: glass melt, refractory
and atmosphere) of front wall and rear wall 1400 o C, that of the
metalline near the hotspot 1500 o C, and that of the palisade block
surface, shows a linear gradient between the above temperatures.
In the above mentioned furnace the total electron equivalents per
hour per mol oxide, with new AZS and a Li 2o-free glass, are:
Bottom area * e.e.h. at 1350 0 C + areas of front and rear walls *
e.e.h. at 1375°C + side wall areas * e.e.h. at 1400 0 C.

e.e.h. of bottom front and side walls total

rear walls

N~O 38.2 12.7 61.7 112.5

K 20 511.3 60.7 195.7 767.6

BaO 59.4 10.8 44.4 114.6

SrO 69.4 11.5 45.6 126.6

The values in the table apply to the above furnace and its
temperatures. The share of K20 is obviously far higher than those
of the other oxides. This difference in the share of the various
oxides implies that the glass composition has a strong effect on
the total e.e.h., thus on the oxygen gas formation potential at
the AZS interface, of a glass melt in an industrial furnace. By
way of example, the table below shows the difference between 354
and 395 glass for the complete furnace.
There is a marked difference between the two glass types, which is
mainly caused by the difference in the potassium content.
 116

glass 354 glass 395 glass

weight% mo1% e.e.h. weight% molt e.e.h.

Na 20 9.35 10.25 1153 6.70 7.56 851

K20 6.45 4.67 3585 9.10 6.77 5197

CaO 1. 50 0.35

BaO 10.70 5.87 673 7.70 3.79 434

MgO 1. 00 0.25

SrO 3.00 1. 96 248 9.50 6.39 809

total 5659 7291

4.4 Discussion.

The strong rise in e.e.h. values using a lithium glass might be

due to the fact that lithium glass has a lower viscosity or to the
interaction between Li 20 and the AZS leading to a rapid decrease
in the AZS glassy phase viscosity due to Li+ penetration. The
difference in the so called melting point (which is by definition
the temperature at which the melt has a viscosity of 10 Pa.S)
between 354 and 395 glass is comparable with that between 395
glass and lithium glass. (The melting point of 354 glass is 1465
°c, that of 395 glass is 1427 °c and that of lithium glass 1400
OC). The e.e.h. values per mol of cation, of 354 and 395 glass are
comparable at identical temperatures, which points out that the
velocity-determining step: the diffusion of glass melt cations in
the glassy phase of the AZS is left rather unaffected by the
difference in composition between 354 glass and 395 glass. The
Li 20 in the lithium glass affects the properties of the glassy
phase in the AZS, in such a way that the diffusion rates of all
the cations are increased (see 3.2.2.4 of this study).
The total summation of e.e.h. values of a glass in the described
 117

model, gives an indication of the amount of oxygen formed by the

diffusion rate controlled mechanism of oxygen formation at the
interface AZS/glass melt.
A glass is capable of absorbing or releasing oxygen, expressed in
moles per unit of time and per unit of AZS surface area. There are
two ways in which a glass can absorb oxygen: physically and
chemically (Lit. 1). When oxygen is dissolved physically, the
oxygen molecules dissolve in the glass melt interstitially, as if
the melt is an inert, more or less open matrix. The oxygen can
also dissolve in the glass by an oxidation reaction of polyvalent
cations in the glass melt. This is called chemical dissolution.
The quantity of oxygen which can be absorbed by the glass melt
depends mainly on the temperature, the redox of the glass and the
concentration of Sb, ee, Fe and S (Lit. 2). For instance, the
equilibrium of the reaction
Sb20 3 + O2 Sbps
shifts to the right hand site when the temperature decreases.
As long as the local potential for oxygen absorbtion by complete
dissolution is equal to or higher than the supply, no oxygen
bubbles will be generated. After the tests with 354 glass at 1350
oc, hardly any bubbles with oxygen were found. It is, therefore,
supposed, as a first approximation, that a TV-screen glass, with
about 0.4 weight% SbP3 and 0.05 weight% Fe 20 H is still capable of
absorbing the oxygen atoms arising from interaction between 354
glass and AZS at 1350 0 C. For the furnace used in the model this
implies an absorption capacity of 4601 e.e.h. (4601 e.e.h is equal
to the total e.e.h. value of 354 glass at 1350 °e, about 25 e.e.h.
per m2 , mUltiplied by the total glass melt/AZS interaction area in
the model).
The potential of glass defect formation owing to interaction
between AZS and 354 glass is (from calculation example 5659 -
4601) 1058 and between AZS and 395 glass is (7291 - 4601) 2690,
which means a glass defect ratio owing to the interaction between
fused cast AZS and 354 glass to 395 glass of 1 to 2.5.
The above calculation is only an estimate, with the hypothesis of
the oxygen absorption capacity having a strong effect on the glass
 118

defect potential ratio. An intensive study to quantify the effects

of oxygen absorption by the glass melt would improve the accuracy
of this estimation. In industrial practice, the mentioned ratio of
glass defects due to the interaction of fused cast AZS, for the
two mentioned glasses (354 glass and 395 glass) corresponds with
the relative amount of glass rejects.
The bubbles originating from the melting end were doubled. The
knots out of the melting end were more then doubled. The sum of
the two glass defects originating from fused cast AZS/glass melt
interaction in the 395 glass was approximately 2.5 times larger
than the sum of both mentioned glass defects when 354 glass was
melted in the same furnace.
Similar observations of increase of glass defects due to fused
cast AZS/glass melt interaction have been made when smaller
increases of K20 content in the glass were applied. Of course, the
larger the increase of K20 content the larger the glass defect
increase.

Generally, in an industrial furnace, there is a temperature

difference over the fused cast AZS block: between the glass melt
touching surface inside and the colder outside surface of the
block. This temperature decrease over the refractory block will
lower the diffusion of cations in the glassy phase in the interior
of the fused cast AZS.
starting with new AZS, the amount of electron equivalents
produced, will drop (slowly) because the chemical composition of
the glassy phase in the AZS and the oxides in the glass melt will
become better in balance.
Previously it has been shown that potassium ions have by far the
highest diffusion rates (due to the larger driving force) among of
the investigated cations at 1350 °C. This means that deep in the
block at temperature levels where the diffusion rates of the other
cations is almost zero, potassium still may have a substantial
diffusion rate. Therefore it will take more time to obtain a
balance (an apparent equilibrium, the reaction kinetics are
stopped by the low mobilities) between the potassium oxide in the
 119

glass melt and the fused cast AZS compared with the other oxides
in the glass melt. In practice a complete chemical balance will
not be reached on account of the AZS corrosion, which increases
the local block temperature. The tests in which the e.e.h. values
have been derived, have been performed under static and isothermal
conditions. The corrosion rates or wear of fused cast AZS in non
static conditions, like in an industrial furnace, are higher than
under static conditions. At the same temperature, a glass melt
with a low viscosity will give a larger corrosion rate (lit. 3-6),
therefore a higher diffusion rate and more production of oxygen.
A rise in temperature will increase the temperature of the
refractory causing a change in the equilibrium of the polyvalent
ions (to the lower valence state) in the fused cast (Lit. 7-9),
thereby releasing oxygen. Also the diffusion of cations will
increase because the apparent equilibrium of the concentration
profiles will shift deeper into the refractory, the difference
between the existing concentration profiles and the apparent
equilibrium concentration profiles of the higher temperature
increases the driving force for the diffusion of cations. At a
drop in temperature the opposite will occur, there will be a
temporary "over-saturation" of the local concentration of the
cations and therefore a decrease in driving force for diffusion.
A furnace in which during a long period, a specific glass
composition has been molten, the fused cast AZS is in "apparent
balance" with this specific glass. In the case of a change in
glass composition the fused cast is out of "balance" with the new
glass composition. As an example, the glasses (354 and 395 glass)
used in this chapter for the e.e.h. calculation, have about the
same total amount of alkali oxides (Na 20 + K20) expressed in
weight%. When for instance a sodium oxide rich glass like 354
glass has been molten for a long time in a furnace and in this
furnace the glass is changed to a potassium rich glass like 395
glass, initially the amount of positive charge by diffusion of
potassium into the fused cast will be partly compensated by the
diffusion out of the fused cast into the glass melt of sodium,
consequently the electron transfer will be relatively small and
 120

therefore also the oxygen bubble formation (comparable with the

bubble formation with the previous 354 glass). The charge
compensation by the sodium diffusion from the glassy phase of the
fused cast AZS will decrease relatively fast because of the
limited amount of sodium oxide in the fused cast AZS. The
diffusion of potassium from the glass melt, however, continues
thereby, increasing the formation of glass defects from the
interface glass melt/AZS. This explains why in industrial practice
the number of glass defects, after such a glass change, is higher
after a few months than in the first few weeks. The diffusion of
sodium in the glassy phase of the fused cast AZS to the glass melt
has been derived in chapter 3, figure 3.2.23 of this study. Under
regular conditions, most of the glass defects arise owing to the
interaction between glass and AZS in the high-temperature areas.
If the AZS interacts with a "high-potassium" glass, the lower-
temperature (1320 - 1350 °C) interaction area (furnace bottom)
may also take a substantial share in the total glass fault rate
(see chapter 6 for examples of industrial furnaces).
Note that not all the defects arising in the glass, end up in the
product or are a cause for rejection.
The chance of survival of bubbles and knots depends on aspects
like their source, their original size, the composition of the
defect, the composition of the glass melt, the temperature of the
glass melt, the redox state of the glass melt, the exchange of
gases from the generated oxygen bubble with the melt, residence
time and glass flow patterns in the melting tank. In the T.V.
screen glass production, it is their ultimate size and location in
the screen which decides whether they are a cause of product
rejection.
 121

Literature references chapter 4.

[ 1] R.H. Doremus;
Diffusion in glasses and melts.
J. of Non-Cryst. Solids 62 (1977) p. 263-292.
[ 2] R.G.C. Beerkens, H. de Waal;
Mechanism of oxygen diffusion in glass melt containing
variable-valence ions.
J. Am. Cer. Soc. 73 (1990) p. 1857-1861.
[ 3] R. Brueckner;
Wechselwirkungen zwischen Glasschmelze und
Feuerfestmaterial.
Glastech. Ber. 53 (1980) p. 77-88.
[ 4] M. Dunkl, Brueckner;
Corrosion of refractory materials by a container glass melt
under the influence of various convection flows.
Glastech. Ber. 62 (1989) p. 10-19.
[ 5] S. Ozgen, E. Aydin, M. Orhon;
The relation between the type of alkali oxide in glass and
refractory corrosion.
Int. Conf. on glass XV (1989) p. 283-286.
[ 6] B. Krabel, R. Brueckner;
Convection flow dependent corrosion of refractory material
in glass melts and comparison to theoretical relations.
Int. ColI. on Refractories XXXVI (1993) p. 45-49.
[ 7] F.W. Kraemer;
Analysis of gases evolved by AZS refractories and by
refractory/glass melt reactions. Techniques and results.
contribution to the bubble forming mechanism of AZS
material.
Glastech. Ber. 65 (1992) p. 93-98.
[ 8] E.L. Swarts;
Bubble generation at glass/refractory interfaces. A review
of fundamental mechanisms and practical considerations.
Glastech. Ber. 65 (1992) p. 87-92.
 122

[ 9) M. Dunkli
Studies on the glassy and reaction phases given off by fused
cast AZS blocks and their effects on the glass quality.
Glastech. Ber. 62 (1989) p. 389-395.
 123

s. Bubble formation during the laboratory-scale tests.

5.1. Introduction.

In the laboratory-scale tests, described in chapter 3 of this

study, the formation of 02-bubbles has been shown.
The objective of the experiments described in this chapter is to
determine the influence of temperature and glass melt composition
on the number of bubbles, their size (diameter) and gas content,
in the laboratory-scale (crucible) tests. The results can be used
as starting parameters for a flow model which includes the
position of bubbles in the glass melt bath and the gas exchange
between bubbles and the glass melt.
The number of bubbles generated and their diameter, in relation to
the temperature, gives an indication of the importance of the
fraction of bubbles from fused cast AZS in the total glass
defects.
Another objective of the experiments described in this chapter is
the investigation of the glass defect potential tendency of a
glass melt/AZS interaction by direct counting the number of
bubbles at the end of a crucible experiment.
Although the original gas content of a bubble can change
dramatically during its residence time in the glass melt (Lit. 1-
8), the gas composition of the relatively fresh bubbles can
confirm the existence of an electrochemical mechanism for the
bubble formation.
Another parameter which has been investigated here, is the effect
of the duration of the interaction between glass melt and fused
cast AZS on the number, size and gas composition of the bubbles.

5.2. Experimental method.

The crucible + glass samples obtained from the tests, described in

chapter 3 'cation diffusion into fused cast AZS' have been cut
into slices, polished and photographed. Next the bubbles were
counted and their diameters measured. The gas composition of a
 124

part (a randomly taken sample) of the generated bubbles in the

crucible have been analyzed.
The gas composition and the volumetric diameter of the bubble have
been analyzed by breaking a bubble in a high vacuum chamber and
analyzing the gas content of the broken bubble with a Balzers QMG
420 quadrupole mass spectrometer. The volumetric diameter has been
determined by measuring the increase in pressure of the known
chamber volume immediately after the breaking of the bubble in the
high vacuum chamber, assuming that the pressure in the bubble at
room temperature is 1/3 bar and the gas of the bubble is an ideal
gas. The diameter of the bubbles is calculated with the equation
P*V/T= constant. The assumption that the gas in the bubble has a
pressure of 1/3 bar has been checked by analyzing numerous
spherical bubbles and is understandable because at about 900 K the
diameter of the bubble is frozen in and further gas exchange is
neglectable.

5.3. Test results.

Annex 4 shows the results of the analysis and annexes 1 and 5

present the compositions of the glass types used.

The number of bubbles, in the glass slices, differs widely from

test to test, also within the duplicate tests. Figure 5.1 shows
the number of bubbles plotted against the temperatures. The
samples tend to contain more bubbles as the temperature increases.
The dispersion at a given temperature is very large. The so called
t-test (annex 6) shows that the average counts at 1350 and 1425°C
differ significantly (t[30]=3.041). The difference between the
number of bubbles at 1425 and 1500°C is not significant
(t[27]=1.198). That between the number at 1350 and 1500°C, again,
is significant (t[19]=3.226).
The results of figure 5.1 show that the method of counting the
number of bubbles at the end of the test is not suitable for
detecting the difference in glass defect potential because of the
large variation of for instance the number of bubbles counted in
 125

duplicate tests (annex 4). Even the influence of the temperature

on the number of bubbles can only be proven statistically after a
large number of tests.

The diameter of the bubbles increases as the temperature rises.

Figure 5.2 shows the average diameters (arithmetic) of the five
largest bubbles found in each test. The average diameter of the
five largest bubbles is described in the figures as the diameter
range. No difference between the different glasses except for
lithium glass has been detected.

Figure 5.1 Number of bubbles Figure 5.2 Average diameter of

per cm3 determined in the the five largest bubbles
crucible experiments with determined in the crucible
different glasses at experiments with different
different temperatures. glasses at different
temperatures.

figure 5.1
Number of bubbles Bubble diameter rangefiQUre 5.2
ER 1681 crucible ER 1681 crucible

bubbles per cm3 Diameter (mm)

70,-------'--"--'---'-'---=--------------, 0.7 ;:-.::=------'----'----'--'----------------,
+ +
60 + + 0.6 +
+
50 + 0.5 + +
+
+
40 + + 0.4 +
+
:j: +
+ +
30 0.3
+
t
:::l ~
20
+

10
+

O~----'--------'-------.J
+
oL +
-'----_ _--'---_ ____'___ _---"'-_ _--.J

1300 1350 1400 1450 1500 1550 1300 1350 1400 1450 1500 1550

Temperature Temperature
average diameter of 5 largeat bubbles in sample.
 126

Remark: The lithium glass is the only glass melt of which the
results of the experiments is slightly different, the number of
bubbles at the end of the test is low and the average diameter
is the largest of all tested glasses.

The t-test results of all the experiments, point to a significant

difference in average diameter between 1350 and 1425°C
(t[30]=5.767), and between 1425 and 1500°C (t[27]=3.988).

Figure 5.3 The relation between Figure 5.4 The relation between
the average diameter of the the average diameter of the
bubbles prepared for gas bubbles prepared for gas
content analysis and the analysis and the average
average diameter of the diameter of the
fivelargest bubbles successfully analyzed
determined in the sample. bUbbles, showing that the
analyzed bubbles are
representative.

figure 5.3 figure 6.4

Bubble diameter Bubble diameter
ER 1681 crucible ER 1681 crucible

average diameter of prepared bubble (mm) average diameter of prepared bubble (mm)
0.6 - 0.6

0.4 + 0.4 +

0.3 0.3
+ + + ++
+
++ +
+ + + + + +
0.2
+
"'i+t+ ++ 0.2
++ +

+ + +
+ \+ ++$-

o~~
+ -<I- +t + +
0.1 + 0.1
+

o L-_-'--_-L.-._--'-_---'--_---'-_--'- _
o 0.1 0.2 0.3 0.4 0.6 0.6 0.7 o 0.1 0.2 0.3 0.4 0.6
bubble diameter range (mm) average diameter of analysed bubble (mm)
 127

Figure 5.3 proves there is a linear relationship between the

average diameter of the five largest bubbles in the sample and the
randomly taken sample of the average diameter of bubbles prepared
for gas composition analyses. The linear relationship indicates
that the largest bubbles, which are large enough to cause reject
in for example TV-screen production are a part of the population
of the analyzed bubbles.
The bubbles prepared for analyses have been taken randomly in the
neighbourhood of the glass/AZS interface. Not all the prepared
bubbles have been successfully analyzed, figure 5.4 shows that
there is no significant difference between the average diameter of
the prepared and the successfully analyzed bubbles.

Figure 5.5 Average (arithmetic) Figure 5.6 Relation between

oxygen content of the the average oxygen content
analyzed bubbles of the of the bubbles and the
tests with different glasses average diameter of the
at different temperatures. analyzed bubbles.

figure 5.5 figure 5.6

Bubbles Bubbles
ER 1681 crucible ER 1681 crucible

Average oxygen content of bubbles Average oxygen content of bubbles

100 -- --------..-.. ----------x7---- , 00 - --------------x-: - - - - -
0' •
90
0
0 90 ,p
,
80 80 0
, , 0
§ 0

70 § 70
, 0
0"

80
" 0
c 80 0 0 0
0

50 0
50
"
0
0 "
0 40 0
40
0
30 0 30

" 20
,,
20

10 / 10L u/
o _ .. ,_-L~_--:-L-
0 " 0 0.1 0.2 0.3 0.4
1300 1350 1400 1450 1500 1550

Temperature Average diameter of analysed bubble (mm)

1350 C 0 1425 C x 1500 C

" 1350 C 0 1425 C x 1500 C "
 128

This justifies the assumption that the gas content of the analyzed
bubbles is representative for the gas content of the randomly
prepared bubbles.

From the analysis it is obvious that the average gas composition

of the bubbles depends on the test temperature and diameter.
The higher the temperature and the larger the diameter, the higher
the oxygen content (figures 5.5 and 5.6). Bubbles from various
tests at 1350 °c hardly contain oxygen gas. At tests at 1500 0 C
there are hardly any bubbles analyzed without oxygen as the main
component.

The effect of the test duration is investigated by experiments in

which the AZS crucible has been exposed longer than 100 hours to
the molten glass, the glass melt has been exchanged several times
during such an experiment. The last renewal of the glass melt has
been performed approximately 100 hours before the end of the test.
In this way in all the tests the residence time of the glass in
the crucible in which the bubbles have been investigated has been
equal. The residence time of the bubbles is much smaller than the
mentioned 100 hours due to their (Stokes) rising speed. The rising
speed of a bubble with a diameter of 0.2 mm is for example 2 cm/h
at melting point temperature (viscosity 10 Pa.s), which means that
such a bubble reaches the surface within 2 hours under these
conditions.
The effect of test duration (the total time the AZS has been
exposed to the molten glass) on the number of bUbbles, present at
the end of the test time, at a specific temperature is not
significant (figure 5.7). Nor has it any significant effect on the
average diameter of the five largest bubbles (range in annex 4) of
the prepared sample (figure 5.8). There is no significant
difference in the average diameters of the bubbles between the
various test duration times (figure 5.9). There is also no
significant difference in average oxygen content on the time scale
between 100 and 260 hours (t[16]=1.156), but there is a
significant difference between 260 and 600 hours (t[ll]=2.106) and
 129

Figure 5.7 - 5.10 The number, the average diameter of the five
largest, the average diameter of the analyzed and the
average oxygen content of the bubbles in relation with the
duration of the experiment.

figure 5.7
Number of bubbles Bubble diameter range';QUr. 5.8
ER 1681 crucible ER 1681 crucible

bubbles per cm3 Diameter (mm)

70,-----'---------------, 0.7,----'----'-----'-------------,

80 x 0.8 +
x
50 x 0.5 +
+ x +
40 + 0.4 -
*~ t+ "
x + "
30
t "
0.3

x +
20
x
x
:\: " 0.2

10 x 0.1

o '--_ +
_'___~--'-____L_ _~_ _'___~_ ____l
o
o 100 200 300 400 500 600 700 o 100 200 300 400 500 600 700
Time (hrs.) Time (hrs.)
test temperature 1425 C test temperature 1425 C

figure 5.9 figure 5.10

Bubble diameter Bubbles
ER 1681 crucible ER 1681 crucible

average diameter of analysed bubble (mm) average oxygen content of bubble (mm)
0.35,---=-----------'-----------, 100 ,-----=---~"'-----------,

0.3 +

0.25
+ 80
t "
t
x "f 60
0.2
+ " "
~ "
0.15
x
1 40 +
X
0.1 X
+
20
0.05

o o
o 100 200 300 400 500 600 700 o 100 200 300 400 500 600 700
Time (hrs.) Time (hrs.>
test temperature 1425 C test temperature 1425 C
 130

between 100 and 600 hours (t[7]=1.989). Although the t-test with
the experiments of a test duration of 600 hours is very disputable
because there are only two measurements of this test duration.
A tendency to obtain higher 02 levels in these bubbles has been
observed for larger test periods (figure 5.10), although also here
the small number of measurements of the experiments with a test
duration of 600 hours makes the observation disputable.

Summarising, it can be concluded from figure 5.1, 5.5 and annex 4

that for the bubbles found in the crucible experiments no relation
between these bubble characteristics and the glass melt
composition could be established.
A temperature increase, increases the number of bubbles, their
diameter and oxygen content in the crucible experiments.

5.4 Discussion

A rise in temperature is accompanied by an increase in the average

bubble diameter and a drop in glass viscosity and consequently, a
reduction in the average time a bubble takes to reach the glass
surface to be released. If the change in viscosity caused by
increasing levels of corrosion products in the glass is ignored
and the glass density is assumed to remain unchanged at the
various temperatures, the ratio of the residence time of a bubble
with the determined average diameter can be calculated. The ratio
between the periods of residence time for a bubble of the average
(constant) diameter in the melt is 11 to 3 to 1 at 1350, 1425 and
1500 o C.
The number of bubbles found in the glass slices, however,
increases as the temperature rises, in spite of the strong
reduction in expected residence time in the glass melt. Thus, the
amount of bubbles formed, at the end of the test duration, is
probably about 25 times higher at 1500 than at 1350 0 C.
This increase can be explained by the results presented in chapter
4 of this study. There are two factors; first the increase of the
diffusion rate of cations from the glass melt into the AZS which
 131

increases the formation of oxygen at the interface glass melt/AZS.

The second important factor is the reduced capability of the glass
melt to absorb oxygen at increasing temperatures, because of the
higher O2 fugacity of the glass melt, mainly caused by the redox
reactions of polyvalent ions (antimony, iron) in the melt
producing oxygen gas at elevated temperatures. The second factor
is also important for the observed decrease in diameter and oxygen
content of the bubbles with decreasing test temperatures. During
the residence time of the bubbles in the glass melt the oxygen
from the bubbles is diffusing into the glass melt. At lower
temperatures, the oxygen fugacity of the melt is larger and thus
the driving force for this diffusion is increased. Also the
average residence time of the bubbles is increased by the larger
viscosity at lower temperatures.
Meden and v.d. Pas (Lit. 9) have determined the average life time
of 100% oxygen bubbles, of bubbles consisting of oxygen and other
gases and of non-oxygen bubbles in similar crucible tests. Their
conclusion also was that the oxygen in the bubbles is absorbed as
the bubble remains longer in the glass at the investigated high
temperatures. Apart from a minor number of bubbles arising, for
example, from pores and cavities in the refractory, most of the
bubbles started as pure oxygen bubbles. The present investigation
shows to the same tendency, although the amount of oxygen bubbles
found after tests at 1350 0 C is too low for the same conclusion to
be definite. At higher temperatures, the number of analyzed oxygen
bubbles is high enough for this conclusion to be fully justified.
However, the wide spread in the results makes the present method
of counting and analyzing bubbles unsuitable as a simple way of
jUdging the tendency for differences in glass defect potential of
glasses with different glass composition.
The dynamic blister test proposed by Dunkle (Lit. 10), in which
with the help of lenses, objectives, mirrors and a video camera,
all the bubbles which are formed in the glass melt in a refractory
crucible are followed in-situ, is probably a better method of
determining the bubble generating potential of a glass
melt/refractory combination.
 132

with this method, measurements of the bubble formation rate at the

fused cast AZS 32 in contact with a glass melt show a decrease in
bubble generation after 260 hours at 1400 °C. These results in the
Dunkl test method agree well with the proposed mechanism and the
measured cation diffusion decrease in time in a crucible
experiment described in chapter 3 of this study.
simply counting the number of bubbles at the end of the test is
only suitable to demonstrate very large differences in the bubble
generation potential.
Summarising, it can be concluded that the applied test method is
suitable to get an insight in the bubble formation mechanism and
the gas exchange of the bubble in the molten glass at different
temperatures. The used test is hardly suitable as a quantitative
method for determining bubble formation tendencies.
The number of generated bubbles, their diameter and gas content
can be used as a starting parameter for a model which calculates
the rise of bUbbles in the glass melt and the gas exchange between
bubbles and glass melt. It is recommended to link such a future
model with a 3-D model of a glass furnace. The result can give a
good estimation of the contribution of the bubbles from fused cast
AZS to the total glass defect number of industrial glass
production. such a future model would be a very powerful tool for
glass quality control in industrial glass production.
 133

Literature references chapter 5.

[ 1] F.W. Kraemer;
Bubble defect diagnosis by means of a mathematical model.
ColI. papers, XIV IntI. Congr. on Glass (1986) p. 288-295.
[ 2] F.W. Kraemer;
Mathematical models of bubbles growth and dissolution in
glass melts.
Gas bubbles in glass; I. Comm. on Glass (1985) p. 92-126.
[ 3] L. Nemec;
The refining of glass melts.
Glass Tech. 15 (1974) p. 153-156.
[ 4] C.H. Greene, D.R. Platts;
Behaviour of bubbles in oxygen and sulfur dioxide in soda
lime glass.
J. Am. Cer. Soc. 52 (1969) p. 106-109.
[ 5] R.G.C. Beerkens;
Chemical equilibrium reactions as driving forces for growth
of gas bubbles during refining.
Glastech. Ber. 63K (1990) p. 222-241.
[ 6] J.M. Hermans, A.C. Verbeek;
Gasaustausch zwischen Blasen und geschmolzenem Glas.
67 Glastechnische Tagung, Koningswinter (1993) p. 34-38.
[ 7] H.O. Mulfinger;
Zum Verhalten von Blasen in Glasschmelzen.
Glastech. Ber. 45 (1972) p. 238-243.
[ 8] H.O. Mulfinger;
Gasanalytische Verfolgung des Lautervorganges im Tiegel und
in der Schmelzwanne.
Glastech. Ber. 49 (1976) p. 232-245.
[ 9] G. Meden, T. v.d. Pas;
Invloed van het uitstoken van ZAC 1681 op de mate van
belvorming (Effect of firing out ZAC 1681 on intensity of
bubble formation).
OC 81/350.
 134

[10] M. Dunkl;
TC-ll meeting (Refractories) in Madrid d.d. 04-10-1992.
 135

6. Knot formation mechanism and characterization.

6.1 Introduction.

A knot is a vitreous particle with a composition deviating from

the composition of the bulk of the glass, and consequently, a
different index of refraction, which brings about a local lens
effect.
The vitreous particle may contain crystals.
Objective of this chapter is to show the relation between the
chemical composition of the knots in the glass products and their
origin in order to find the source of these important glass
defects. In T.V. screen glass production, for instance, a product
with a knot with a diameter of 1 mm. is in most cases a reject.
A knot, originating from fused cast AZS in TV-screen glass has a
higher AIZO)-content than the bulk glass, which leads to a
viscosity and surface energy of the knot higher than for the bulk
glass. The density of these knots, however, is lower than that of
the bulk (TV-screen) glass.
In the literature review of chapter 2 the gas formation (bubbles)
in the refractory and at the interface refractory/glass melt is
found to be the most important cause of pushing liquid phase out
(exudation) of the fused cast AZS.
This means that, the mechanism and rate of bubble formation
investigated in the previous chapters of this study, is also
determining for the knot formation at the interface glass
melt/fused cast AZS.
For the elimination of a knot defect problem in the practice of
industrial glass production, an indication of the origin of the
knot in the furnace is very helpful.
In this chapter, the chemical composition of knots in products
fabricated in the past will be investigated. The dissolving of
knots will be investigated and the preservation and/or change of
the chemical composition of the knot during its residence time in
the glass melt. The effect of the local temperature and the
chemical composition of the glass melt on the chemical composition
 136

of the interface glass meltjAZS is also studied in this chapter.

By combining information on the chemical composition of the knots
with data on the chemical composition of the glass meltjAZS
interface obtained from laboratory experiments, the last part of
the chapter will show the possibility to link a production knot
defect problem to its origin temperature area where it has been
formed in the furnace.

6.2 Knots in TV-screen glass products.

Of the knots in products fabricated in the past few years the

chemical composition of the glassy phase, the location in the
screen, the size and the kind of crystals (if present) they
contain have been determined. No relation could be established
between the knots location and size, chemical composition or
containing crystals. Nor between the knots size and location,
chemical composition or containing crystals. However, a relation
has been found between the Al 20 3 content of the glassy phase of the
knot and, for instance, the content of BaO, Si02 , K20 and Zr02 of
the glassy phase of the knot.
The applied analysis is identical with the SEMjEDX method
described in chapter 3, part 3.2.1.1. Only the sample preparation
is different, a straightforward polishing treatment of the knot is
applied.
The concentration of Al 2 0 3 of a large number of analyzed knots in
the TV-screens plotted in a graph, against the concentration of
one of the other components, for instance BaO, Si02 , K20 and Zr02 ,
has a distinct shape.
Figure 6.1.a to 6.1.d show the Al 20 3 related to measured BaO
concentrations (in weight%) of knots in TV-screens. The knots in
the TV-screens from an industrial Philips furnace PTA plotted in a
graph (figure 6.1.a), having the distinct shape of a crescent with
tips up. The knots in the TV-screens from furnace PBA (figure
6.1.b) have a similar shape. The knots in TV-screens from furnace
PAB (figure 6.1.c) are almost exclusively high-Al 20 3 ones (more
than 15 wt%), which means that only half of the crescent is
 137

Figure 6.1.a The Baa versus A1 20 J concentration (in weight%) of

knots, of furnace PTA, in TV-screen glass.
CHEMICAL COMPOSITION OF KNOTS IN PRODUCTS
0
17 - ..
<I
or:>
~
+
....
:I: 10
+
<!>
H
UJ
:3

+
.........................
} +
Furnace
. . #:.
+,.. + +
+.t:+ t
.....................f.~ ... ~ ...+. .. +
+
++ *t ++
..
+
+
....... -. : +•.... ····tt··:".···········.. ····
t+ t+ t ~ + ++ +
+ + +
+ + +
+

+----~----~------,-----TI-----'------"

o 5 10 15 70 75 30
WEIGHT X AL203
Figure 6.1.b The Baa versus A1 20 J concentration (in weight%) of
knots, of furnace PBA, in TV-screen glass.
CHEMICAL COMPOSITION OF KNOTS IN PRODUCTS
0 17
<I
or:>
~

....
:I: 10 ,.,
<!>
H
UJ
:3
D
D
8 .... ... ~.

Furnace
D
D
b .....

I " I
o 10 15 20 25 30
WEIGHT X AL203
 138

Figure 6.1.c The BaO versus A1 20 3 concentration (in weight%) of

knots, of furnace PAB, in TV-screen glass.
CHEMICAL COMPOSITION OF KNOTS IN PRODUCTS
a 12
<I:
l%:l
~ x

~ 10
C>
.....
w
3 x

8 ··········x············~·~···
x x
" )( )( ,,*ll'
x~T
x
xX
Furnace
)()( \)(~ X xx
b uXu·xuuu~";4xullc"x,(u.~x.uu;/',,.u. ur x PAB I
Xx if x~",,~Wxx Xx ----
)( )( ~)()( )(
)( x x x
xxx
x xxx
................................ ...........x x x........ x
x x xx
x x x
x
x x
"
2 .

0~,-----.----.-----,---r-1 ---,-1-----,
o 10 15 20 25 30
WEIGHT r. AL203
Figure 6.1.d The BaO versus A1 20 3 concentration (in weight%) of
knots, of furnace PSA, in TV-screen glass.
CHEMICAL COMPOSITION OF KNOTS IN PRODUCTS
a 12
<I:
l%:l
~

t-
:r 10·
C>

.. ....
H
W
3

.. .. Furnace

..,'\t.~ •
••• o·
. ··t··.·········
~.

....
~.6 •

o+-----~-~---,-­ ,. l
10 15 25 30
WEIGHT r. AL203
 139

visible. The knots in the screens from furnace PSA (figure 6.1.d)
had a few incidents as source (calamities). All but one of these
analyzed knots in the PSA products have a low AI 20)-content (less
than 15 wt%), again only half of the crescent shows up.
All the knots from all the furnaces together form the whole
crescent again, despite the difference in glass type melted in
these furnaces.
Table 6.1 shows that the variation in BaO-contents of these
glasses is rather wide.

Table 6.1

The main components of the glasses molten in the different

furnaces in weight% (nominal values).

furnace glass type Si0 2 AI 2O) Na 20 Kp BaO

PTA 328 66.1 2.9 7.0 6.4 12.3

PBA 346 63.8 3.3 9.1 6.5 8.7

PAB 354 glass 63.9 3.3 9.5 6.5 10.7

PSA 346 63.8 3.3 9.1 6.5 8.7

Figure 6.2 shows the linear relationship between the

concentrations (in weight%) of AI 20) and Si02 in the knots.
Figure 6.3 showing the K20-content against the AI 20)-content in a
large number of analyzed knots yields a crescent as in the case of
BaO, although in the present case one with tips down.
All the "pure" glasses in table 6.1 are free from Zr0 21 although
the knots do contain this component (figure 6.4).
Apart from some parts of the refractory in the forehearth, AZS is
the only furnace material which contains Zr0 2 • Thus this is a very
strong indication that the AZS is the source of the knots.
Figure 6.4 also shows that high zro 2 -concentrations (> 4 wt%) are
most likely to occur when the AI 20)-content is low « 20 wt%).
 140

Figure 6.2 The Sio 2 versus Al 2 0 3 concentration (in weight%) of

knots in TV-screen glass.
CHEMICAL COMPOSITION OF KNOTS IN PRODUCTS
c-<
a
70 " .. ..
+
H
(I)
D
~

I-
b5 p ..

::x:
C>
H
UJ
:3
b0 Furnace
x PAB
55
PTA
.
+

PSA
50
D PBA

45

::J I
10
I
15
I
20 25 30
I

WEIGHT % AL203
Figure 6.3 The Kp versus Al 2 0 3 concentration (in weight%) of
knots in TV-screen glass.
CHEMICAL COMPOSITION OF KNOTS IN PRODUCTS
o 12
c-<
""
~
+
+ + D
I-
::x:
C>
H

~ 10

D
b .. ~.

4+---,-----r----,-------,,-------,------,
o 5 10 15 20 25 30
WEIGHT % AL203
 141

Figure 6.4 The Zr02 versus Al 20 3 concentration (in weight%) of

knots in TV-screen glass.
CHEMICAL COMPOSITION OF KNOTS IN PRODUCTS
N 10
0
""
N
><
r- D
:t:
<.!l 8
.....
UJ
x
:3 x

b
•• +.-
................t •... ,....
D

D
x
Furnace
PAD
+ .,'" + )(
+~f t .9 + PTA
+ t+a++~
...4' \o.:~~o D •
PSA
+ +
...................t." .............• '. 0
+
... ~

<I" • •
a 0
0
.°1
0 °,.0

2
•
D
D
0
0 10 15 20 25 30
WEIGHT " AL203

The solubility of Zr02 in the glass increases when the Al 20 3 -

content decrease which is also reported by Manfredo and McNally
(Lit. 1).
Often the knots contain inclusions, the most common sort of
inclusion is recrystallized Zr02"

6.3 Dissolution of knots

Tests have been carried out introducing synthetic knots in glass

melts in order to investigate the knot dissolution process. The
knots in the bulk of the glass show a different dissolution
behaviour than the knots at the glass surface.
 142

Dissolution of knots in the bulk of the glass

The diffusion profiles remain relatively stable on the time scale.

The knot ascends in the molten glass, because its density is lower
than that of the bulk glass. Besides, its viscosity increases
strongly as the Al 20 3-content increases. The ascending velocity of
the knot in the molten glass raises drag forces at the border of
the knot. The lower the viscosity of the material the larger the
transport of this material due to the drag forces. The combination
of the ascending knot with an increasing viscosity at the border
going to the inside of the knot, yield an effective diffusion rate
that increases as the Al 20 3-content in the concentration profile
drops. The result is that instead of a flattening of the
concentration profiles on the time scale, a kind of balance arises
with stability of the slope of the diffusion profiles on the time
scale (figure 6.5). The diameter of the knot (diameter with
constant high AI 20] content) of course, decreases on the time
scale.

Dissolution of knots at the glass surface

The lower density of the knots causes them to rise to the glass
surface. Owing to the higher surface tension of the knot glass
(0.350 N/m against 0.290 N/m at 1400 0 C) the bulk glass melt tends
to cover the knot glass. This phenomenon is similar with the
'Marangoni effect', which occurs at the three phase contact point:
glass melt, AZS and atmosphere (metalline). The described
phenomenon accelerates the dissolution of the knots in a similar
way as the Marangoni effect accelerates the metalline corrosion
(Lit. 2).
The difference in dissolution behaviour is shown in figure 6.6 for
the knots dissolving in the bulk of the glass and in figure 6.7
for those dissolving at the glass surface.
Figure 6.8 shows the results of the analysis of the concentration
profiles of a knot dissolving at the surface.
 143

Figure 6.5 Concentration profiles for different components in

synthetic knots after different periods of time, at
1350 °c, present in the bulk of 354-glass (the
synthetic knots have different original diameters).

SYNTHETIC KNOT DISSOLVING FOR 1 HOUR.

AT 1350 C IN THE BULK OF THE GLASS.
~
30
I-
:J:
'"
I-<
UJ
:3 25

70 __ al203
__ zro2
J5 __ k20

-e-- bao

J0

o-1---........-....-~:--_--,-------------T------------T··· -·1
o 500 J000 J~)00 7000 7500
MICROMETERS
SYNTHETIC KNOT DISSOLVING FOR 8 HOURS.
AT 1350 C IN THE BULK OF THE GLASS.
30
~
T
I-
:J:
'"
I-<
UJ
:3 25

__ al203

- - zro2
J5 __ k20

-e-- bao

J0

o-+---~,,¥=----,-------,----------,--_. -----1
o 500 J000 1500 2000 2500 3000
MICROMETERS
 144

Figure 6.6 Photographs of synthetic knots dissolving for

different time periods, at 1350 °C, in the bulk
of the glass.

Dissolution of a knot in the bulk,

of the glass for 2 hours at 1350 °C.

Dissolution of a knot in the bulk,

of the glass for 4 hours at 1350 °C.
 145

Figure 6.7 Photographs of synthetic knots dissolving for

different time periods, at 1350 °C, at the glass
surface (vertical cut).

Dissolution of a knot at the glass

surface for one hour at 1350 °C.

Dissolution of a knot at the glass

surface for two hour at 1350 °C.
 146

Both in the bulk of the glass and at the surface the chemical
composition of the knot remains stable until the slopes of the
concentration profiles meet after a certain time period (figure
6.8). For example, in figure 6.8 the slopes of the concentration
profiles at 1700 micrometer meet and the concentration does not
remain stable, the slopes of the concentration profiles at around
1150 micrometer do not meet and therefore the chemical composition
of the knot material between the two concentration profiles is
still equal to the original chemical composition.
These results are in agreement with the investigations of Uemura
and Tabuchi (Lit. 3).
Even tests at 1425 instead of 1350 0 C did not yield any change in
chemical composition of the cores of the knots.

Figure 6.8

SYNTHETIC KNOT DISSOLVING FOR 2 HOURS.

AT 1350 C AT THE SURFACE OF THE GLASS.
...
l-
')~T--

x<!l
H
W
:3 i ~,

l~
___ ill;3!
--+-- zro2

1" k2o-j
-e- bao

J~

~, - --

~ - ~--~~-_!:!'!---";--T":'----':"'----_:":'----~----~-
~__~_'!"__~_':"o:_~_ -"""1""------------1
~ ~J0~ 1000 1 ~,~~ 100~ 2500
MICROMETERS
 147

Table 6.2 shows the results of the SEM/EDX analyses of the

chemical composition of the core of the knot under various
conditions.
The chemical composition is not affected by either time or
temperature. If the composition of a knot remains stable until the
slopes of the concentration profiles meet, the knots in the
products have the same chemical core composition compared to their
original state (at least, as long as the slopes of the
concentration profiles do not meet) .
The preservation of the chemical composition of the knot is a very
important condition for the identification method of the source of
the knot later in this chapter.

Table 6.2 The chemical composition of the core of a synthetic knot

after experiments investigating the knot dissolution
process in molten glass.

tempe- time sio 2 Al 20 3 K20 BaO

rature (in in in in in
hours) weight% weight% weight% weight%

glass 63.85 3.3 6.45 10.7

1350 0 C 1 knot 47.42 26.55 8.96 6.07

2 knot 47.07 26.60 9.06 6.12

4 knot 46.75 26.38 9.44 6.38

6 knot 46.71 26.37 8.97 6.54

8 knot 47.58 26.36 8.88 6.37

1425°C 1 knot 46.86 26.48 8.96 6.48

2 knot 46.93 26.69 8.96 6.30

 148

6.4 Concentration profiles of AZS-to-glass interface

Chapter 3 'Cation diffusion into fused cast AZS' shows the

concentration profiles of NazO, KzO, BaO and SrO at the AZS/glass
melt interface after a test. The physical properties of the
interface and the knots arising from it are strongly affected by
the Al z0 3 content. The interface, in this case, is defined as the
layer stretching from the glass end at the point where the Al z0 3 -
content starts to increase until the AZS end where it reaches its
maximum in the glassy phase.
Figure 6.9 shows the effect of convection at the various
temperatures, with 'w' is the velocity at which a rod (diameter
21 mm) revolves in a crucible with an inner diameter of 70 rom. The
forced convection has hardly any effect on the Al z0 3 concentration
profile in the defined interface.
Figure 6.10 shows the effect of the test time on the interface
concentration profiles at the various temperatures. The schedule
of glass renewal has been the same as described in 3.2 of chapter
3 in this study. After 100 hours, further exposure no longer has a
large effect on the defined interface itself. The temperature,
however, does affect the interface profiles. Higher temperatures
result in more penetration of the profile into the AZS and gives
slightly higher Al z0 3 concentration peaks.
All the graphs are those of the concentration profiles on the
inner crucible side wall (glass side).
Figure 6.11 shows the difference of the Al z0 3 concentration
profiles between the side wall and bottom location. The interface
layer on the bottom is about half as thick as that on the side
wall due to the lower density of the interface layer compared to
the glass melt. At 1350°C, the interface layer on the side wall is
about 1 mm thick and on the bottom about 0.5 mm. In both cases
(side wall and bottom) the interface layer is thick enough for the
formation of knots which have the same diameter. Annex 2, table 6
shows the concentration profiles in a table.
 149

Figure 6.9 The effect of forced convection on the chemical

composition of the interface glass melt/fused cast
AZS 32/33, at various temperatures.

Concentration profiles of A1203.

354 glass, 100 hrs. melting time.
1350 C
40 AJ203 weight %

35
30
AZS glass
25
20
15
10
5
o
-10 -8 -6 -4 -2 0 2 4 6 8
mm

~- w = 0.0 -+-- W = 0.2 ~- w· 2.0

Concentration profiles of A1203.

354 glass. 100 hrs. melting time.
1425 C
40 AI203 weight % '-;-r~~---------,
35
30
~~\'
~-"~~-
25 AZS ..------------- glass

20
15·
10
5
o ~~~~~-~----'-----'------'-.------'----~

-10 -8 -6 -4 -2 0 2 4 6 8
mm

--<>-- w = 0.0 --+- W • 0,2 -~- w • 2.0

Concentration profiles of A1203.

354 glass, 100 hrs. melting time.
1500 C
40 AI203 weight % . -_ _

35
30
AZS glass
25
20
15
10
5
O'-----~'--------'---- _----'_-----.J'--------------.L---.-__ ---.1 ---.L- _ _

-10 -8 -6 -4 -2 o 2 4 6 8
mm

--<>-- w • 0.0 - .... w • 0.2 ~ w • 2.0

 150

Figure 6.10 The effect of test duration on the chemical

composition change of the interface glass melt/fused
cast AZS 32/33, at various temperatures.

Concentration profiles of A1203.

395 glass, 1350 C.

AI203 weight %
40,------=--------,----------,
35
30
glass
25
20
15
10
5
o
-10 -8 -6 -4 -2 o 2 4 6 8
mm

~ 100 hrs. -+- 260 hrs. ~ 600 hrs.

Concentration profiles of A1203.

395 glass, 1425 C.

40 AI203 weight %
35 ~~-
30 ----------
25 ,,----------
20 AZS glass
. 15
10
5
o L L-------.L _ ________ L __-------.l-. --.L_

-10 -8 -6 -4 -2 o 2 4 6 8
mm

-~- 100 hrs. 260 hrs. -~ 600 hrs.

Concentration profiles of A1203.

395 glass, 1500 C.

AI203 weight %
4 0 - - - - - -.----
35
30 glass
AZS
25
20
15
10· ------~
----+
5
o
-10 -8 -6 -4 -2 o 2 4 6 8
mm

-" 100 hrs. -+- - 260 hrs.

 151

Figure 6.11 The difference, of the interface glass melt/fused

cast AZS 32/33, between side wall and bottom.

concentration profiles of A1203.

395 glass, 1350 C.

AI203 weight %
40,------=--------,----
35
30
glass
25
20
15
10
5
O'-~'-----

-10 -8 -6 -4 -2 o 2 468
mm

-e--- side wall bottom

600 hrs.

Concentration profiles of A1203.

395 glass, 1350 C.

AI203 weight %
40 ,------=-------,---------~
35~~----~---.
30 .........\
AZS ~ glass

~~ ~,
::o'------ ---L
\~
-l_~==:=L1=~=:::'I
-1000 -500 o 500 1000
micrometer

--a-- side wall -+- bottom

600 hrs.
 152

6.5 Chemical composition of the interface glass melt/fused cast

AZS.

The curves presenting the BaO concentrations versus AI 20]

concentrations (in weight%), of the knots in products, in figure
6.1 is crescent-shaped.
with the results of measurements which have been made, of the
glass/AZS interface composition going from low AI 20] (glass side)
to high AI 20] concentration (AZS side), of the crucible samples,
plotted in the same way, the crescent shows up again (figure
6.12) .

figure 6.12

CHEMICAL COMPOSITION OF TRANSITION LAYER AZS-GLASS.

)( 354 GlASS NJT REfRESHED. f£LlING TM 1" HOURS.
+ 354 GLASS REFREstlED TWICE. ttEl TOC 1M 2411 HelMS.

<> 12 • 3t5 StASS P«JT Il£FRESHED,. HELTDe TItlE 111 HOURS.

~ a 315 RASS REfM:SH£D lWIC£. t£LTIHG T:DIE ttl. HOJRS.

,. .
.
1425 C
'
~
&

o -----.
B 5 10 15 2B 25 3B
WEIGHT X Al.2D3

The position of this crescent depends (figure 6.12) on the

composition of the glass, but in the case of BaO, not on time.
The crescent-shaped curves of the K20 versus AI 20] concentrations
(in weight%) of the knots in the products (figure 6.3) can be
derived from the chemical composition of the interface (figure
6.13) in a similar way. In this case, however, it is not only the
glass composition that affects the position of the crescent, also
the test duration and number of glass renewals.
 153

Figure 6.13

CHEMICAL COMPOSITION OF TRANSITION LAYER AZS-GLASS.

lC J!Ji4 5lASS NIT RERESHED. tEllING 1M 111 HOURS.
3~4 R.ASS RUIlESHED TWICE. I'ElTIHG 1M 2'" HOURS.

~ 12 3.5 GlASS tIlT R£fRESHED. tEllING 111'£ 111 HOORS.

CI 395 SLASS R£FItESH£D lVItE. t£LTItC 1M n• ..:JURS.

DD

DD

1425 C
~D

4+----,----r--------r,-----,----,-----,
8 10 15 28 25 38
WEIGHT X Al203

The glass-to-crucible interface, of experiments with the glass

(354 or 395 glass) twice renewed and a total melting time of 260
hours contains more K20 with identical A1 20 3-concentrations,
compared to the experiments with a total melting time of 100 hours
without glass renewal.
Figure 6.14, too, shows the extent to which the BaO-to-Al 20 3 ratio
is stable for a specific glass type (395 glass, annex 1), at a
specific temperature, on the time scale. These graphs also show
that the crescent remains stable even with a total melting time of
600 hours and six renewals.
In chapter 3 in this study, it was shown that the BaO-content in
the interface increases as the temperature rises. The same effect
shows up with the BaO-content of the interface plotted against the
Al 20 3 -content. Figure 6.15 and 6.16 show the dependence of the
interface composition on the temperature for 354 glass and 395
glass (Annex 1). The chemical composition of the interface agrees
with that of the original composition of the knots and most often
with the final composition of the central part of the larger
 154

Figure 6.14 The concentration of BaO versus Al 20 3 of the interface

glass melt/fused cast AZS 32/33, at various
temperatures and test durations.
CHEMICAL COMPOSITION OF TRANSITION LAYER AZS-GLASS.
395 GlASS NJT REFRESHED. t£lTING lItE 111 HOURS.
lIS GLASS REFRESHED TWICE. I'£l TIMi TM nl HOORS.
• 395 GLASS REFRESHED SIX TItlEs. '€lTlNG 1M bl' HOlJRS.

1350 C

.... .'
...
~

.. • ••+ ++ .. -w . .'

0+------,-------,------,-------,----,------,
o 10 15 10 15 30
WEIGHT % AL20J

CHEMICAL COMPOSITION OF TRANSITION LAYER AZS-GLASS.

315 GlASS NJT REFRESHED. HELTH 1M 111 HOURS.
395 GlASS R£FR£SHED TWICE. tlELTItC 1M ~a.. HOlM.

a: 11 395 GlASS REFRESHED SIX TD£S. HELTDC TJt£ ... HlJ.RS.

1415 C

'...
I
....
:: ..... x. w+.x ...... , •
• + ••

0+-----.-----,------,-------,------,---,
o 10 15 15 30
WEIGHT % AL20J

CHEMICAL COMPOSITION OF TRANSITION LAYER AZS-GLASS.

395 GLASS 1C11 l6::Al:ESHED. tEl.TIMi 111'l£ 111 HOURS.
315 GlASS REFRESHED TWIet'. t£lTINli TIHE: 2It1 HOURS.

1501 C

.'
'.

,+----,----,------.---,-------.-----,
o 10 15 15 38
WEIGHT X AL10J
 155

Figure 6.15 and 6.16 The dependence, of the concentration of BaO

versus A1 20 3 , of the interface glass
melt/fused cast AZS on the temperature.

figure 6.15
CHEMICAL COMPOSITION OF TRANSITION LAYER AZS-GLASS.
x 354 GLASS. MELTING TIME 100 HOURS AT 1350 C.
+ 354 GLASS. MELTING TIME 100 HOURS AT 1425 C.
o 12
ai + 354 GLASS. MELTING TIME 100 HOURS AT 1500 C.
~

!i: 10 .................................~1: .
~ x ~
H +;-.
~

~
8 x. . \ ..

+
b·······.. · ....·· ••••
4 .

2 .

0--j-----,-------, ---,- -.- -.-_ _- .

o 5 10 15 20 25 30
WEIGHT X AL203

figure 6.16
CHEMICAL COMPOSITION OF TRANSITION LAYER AZS-GLASS.
x 395 GLASS. MELTING TIME 100 HOURS AT 1350 C.
+ 395 GLASS. MELTING TIME 100 HOURS AT 1425 C.
12 ..
~ + 395 GLASS. MELTING TIME 100 HOURS AT 1500 C.

J ?' +
8 ....•.•. ":'J ..... :>.."'.....
xX +
\ott
b ..................~
+
+
#. . ........ ~ ?
+.
.....~
~~.
4
• x R --
2 .

0+--------r-----,-----,----,-------,-----,
o 5 10 15 20 25 30
WEIGHT X AL203
 156

Figure 6.17

A bubble in the interface squeezes a part of the

interface layer, which contains Zr02 crystals,
into the bulk of the glass.
 157

knots which are mainly responsible for the reject of the glass
product. As described in previous chapters, cation diffusion in
the AZS causes oxygen bubble formation in the interface.
A bubble in this interface can force an interface particle into
the bulk glass (figure 6.17) (Lit. 4 and 5). On account of its
high surface tension (due to the high Al 20 3 concentration compared
to the bulk of the glass), this particle becomes spherical. This
means the formation of such a bubble can be accompanied with the
formation of a knot.

6.6 Examples of knots in industrial TV-screen glass production

and their s~urces.

The combination between the stability in chemical composition of

the knot core during dissolution, resulting in preservation of the
original composition of the glass-to-AZS interface, and the
temperature dependence of BaO versus Al 20 3 in this interface,
allows an estimate of the temperature at which the knot has been
formed.
Figure 6.18 shows clear knots (without crystals) found in glass
products from furnace PAB arising in the first year after the
overhaul. A comparison between the position of the BaO versus Al 20 3
concentration of the knots in figure 6.18 and the curves in figure
6.15 proves that most of the knots have arisen between 1425 and
1500°C, which points to palisade blocks in the melting end as the
most likely source.
Figure 6.19 shows the knots arising during the 395 glass-glass run
in PAC. A comparison of their location with the curves in figure
6.16 proves that the knots have arisen around 1350°C, which
probably points to the colder bottom as the source.

Chapter 4 'Effect of glass composition and temperature on glass

defect potential originating from refractory' shows that, at
bottom temperatures (1300 - 1380 °C), oxygen bubble formation
accelerates by the mechanism described for a glass with a high
K20-content.
 158

Figure 6.18 and 6.19 Examples of the A1 20 3 against BaO

concentration in TV-screens.

figure 6.18
CHEMICAL COMPOSITION OF CLEAR KNOTS IN PRODUCTS
DURING START UP OF PAB FURNACE WITH 354 GLASS.

.• ·····lil····

. - . -...
•• T •.
'~

., .. ..
" "ii'···· ···ii···..•···········•···•······ , ..

1-
.• ~::~:.~.

. .~
•

~ - - - - - -..·T --.... ----....-- --T

I I I
~ J~ J5 25 3~
WEIGHT X AL203

figure 6.19
CHEMICAL COMPOSITION OF KNOTS IN PRODUCTS
DURING 395 GLASS RUN IN PAC FURNACE.
L
..,<r
CJ

x
I-
::c 1 ~
t!>
H
W
:3

..• ..
4 ..
•• •
•. ~."
.
. ,. .;.. .- ...
...
• -.
. -

L- . •

········T -T - - - - - - , - - ·------1
J~ L~ 25 3~
WEIGHT X AL203
 159

The K20 content is far higher in 395 glass (9.1 weight%) than in
354 glass (6.5 weight%), this results in hardly any knots arising
from the bottom by this mechanism for the case of 354 glass and
far more knots are generated in the 395 glass which is in close
agreement with the knot analysis. The absence of knots originating
from higher temperature areas in figure 6.19 does not mean that no
knots have been formed in high temperature areas. The knots in
products are always a combination of formation rate and chance of
survival in the glass melt. In the example of the 395 glass the
lower formation rate at bottom temperatures is compensated by a
high chance of survival, which points to a source close to the
throat of the melting end.
In the beginning of the present chapter it has already been
pointed out that the knots in glass from the PSA furnace (figure
6.1 to 6.4) are originating from a few specific incidents
(calamities): namely glass level fluctuations. The presence of
about 10 weight% AIJO J and a high (4-5 weight%) Zr02 concentration,
indicates that the origin of the knots comes from above glass
level (metalline, silica superstructure). Maximum Al 20 3 -contents in
knots originating from the bottom of the PAC furnace are on
average about 20 weight% and the knots originating from the
palisade blocks of the PAB furnace have an average maximum Al 20 3
concentration in the knots of about 25 weight%. These knots with
their sources are frequently observed for TV screen glass
producing furnaces.
If no temperature curves of a specific glass type or any similar
type are available, the A1 20 3 -content may provide a very rough
indication of the potential source in the tank, for knots
originating from below glass level in TV glass melts (remark:
knots originating from refractories of the superstructure (above
the glass level) are not the sUbject of this study).
 160

Literature references chapter 6.

[ 1] L.J Manfredo, R.N. McNally;

The Corrosion resistance of high-Zr02 fusion-cast
A~03-Zr02-Si02 glass refractories in soda-lime glass.

J. of Mat. Science 19 (1984) p. 1272-1276.

[ 2] H. Jebsen-Marwedel;
Dynaktive Fltissigkeitspaare.
Glastech. Ber. 29 (1956) p. 233-238.
[ 3] H. Uemura, H. Tabuchi;
Characterization of typical knots in window glass.
J. of non-Cryst. Solids 38 & 39 (1980) p. 791-796.
[ 4] M. Dunkl;
Studies on the glassy and reaction phases given off by
fused-cast AZS blocks and their effects on glass quality.
Glastech. Ber. 62 (1989) p. 389-395.
[ 5) D. Walrod;
A study of the driving force behind AZS glassy phase
exudation.
Cer. Eng. Science Proc. 10 (1989) p. 338-347.
 161

Annex 1

Composition and properties of investigated glasses.

Glass Glass Glass

Chemical composition 354 395 lithium

SiOz 63,85 weight % 62,06 weight % 65,86

Al z0 3 3,33 0,80 4,96
LizO 3,84
NazO 9,43 6,84 3,30
KzO 6,46 9,10 3,72
MgO 1,05 0,28
CaO 1,48 0,37 0,96
SrO 3,01 9,59 8,25
BaO 10,75 7,70 8,34
ZrO z 2,10

Properties

Density in kg/m3 at 2679 kg/m 3 2750 kg/m3 2672

room temperature
Viscosity:
strain point ( 10 13 .5 Pa.S) 474 °c 497 °c 451
annealing point (10 12 Pa.S) 507 528 481
softening point (10 66 Pa.S) 685 706 649
working point ( 10 3 Pa.5) 1015 1017 958
melting point (10 Pa.5) 1465 1427 1400

Constants Fulcher equation: 10 log mu = A + (B/(T - C) )

mu = viscosity in dPaS
T = temperature in °c
A -1,756 -2,064 -1,638
B 4879,317 5055,011 4522,540
C 165,184 183,028 155,642
 162

Annex 2.

In this annex the large amount of measurements concerning the

concentration profiles is presented in figures and tables.
As explained in chapter 3, every concentration profile of an oxide
consists of about 50 spot measurements.
When the spot measurements are individually printed, like in
figure 2.1 it can become very confusing when several concentration
profiles are shown in one picture.

figure A 2.1
Experiment 127
Na20 diffusion
Na20 weight %
16

14
AZS glass
12 ...

10 ..
6
ill
... t" ~ • • ill t .. ,
6
..
......

0
-8 -7 -6 -5 -4 -3 -2 -1 0 2 3 4 5 6
mm

.. 39,5 V
applied external voltage
1425 C. 354 glass, after 48 hours
800 mm2 electrode area

Also the data in a table become very comprehensive.

To concentrate the essential information and increase the
surveyability, the measurement spots of figure A 2.1 are
transformed to a figure with straight lines, like figure A 2.2.
Such a figure can also be brought more easily into a table
presentation.
 163

Explanation of tables.

experiment number of the test.

glass used glass (cullet).
time test duration.
voltage applied external potential.
surface the surface of the electrode in the glass melt.
cation oxide of cation under investigation.
start pt. the measurement starts in the glass at 6000
micrometer from the first zirconia nodules, the
start point (start pt.) is the distance from the
first zirconia nodules where the concentration
changes.
C bulk concentration of the oxide between 6000
micrometer and start point in weight%.
C top/bot concentration of oxide at maximum or minimum
peaks in weight%.
distance distance in micrometer between C bulk and C
top/bot or between C top/bot and the next C
top/bot.
cum. dist. cumulative distance which gives the position in
the sample of the measured concentration with
regard to the first zirconia nodules.

An example explains the tables: Figure A 2.2 gives the

concentration profile of the Na 20 of experiment 127. In table 1
this is the fifth row.
Experiment 127; the glass used was 354 glass, the duration of the
experiment 48 hours, the applied external voltage 39.5 V with a
glass electrode with a surface of 800 mm2. The concentration
profile of Na 20 is shown in figure A 2.2. The measurement starts
at 6000 micrometer from the first Zr02 nodules of the interface
glass melt/AZS in the glass, until 750 micrometer (start pt.
micro) from the Zr02 nodules a stable concentration of 7.48
weight% (C bulk wght%). A drop in the concentration down to 6.69
weight% (C top/bot wght%), the slope in the concentration decrease
is 350 micrometer (distance micro) long. The concentration 6.69
weight% is reached at 400 micrometer (cum. dist. micro) from the
 164

ifirst Zr0 2 nodules. An increase in the concentration to 8.50

/weight% (C top/bot wght%) across 550 micrometer (distance micro),
the 8.50 weight% is reached at -150 micrometer (cum. dist. micro),
150 micrometer in the AZS material. Then a concentration drop
again to 4.62 weight% (C top/bot wght%) across a distance of 3850
micrometer (distance micro), the 4.62 weight% Na20 is reached at -
4000 micrometer (cum. dist. micro), 4000 micrometer in the AZS
material. The final increase in concentration to 13.99 weight% (C
top/bot wght%) has been across 4000 micrometer (distance micro)
and the measurement has been done at 8000 micrometer (cum. dist.
micro) in the AZS material. All measurements in the fused cast AZS
have been done in the glassy phase.

figure A 2.2
Experiment 127
Na20 diffusion
Na20 weight 'II>
18 r---~-----~--------.~-------------,

2
O'-----'-_---'-_---'--_-'-_-.L-_L--.-J_--'._---L_--'-_--L-_.J.-_L--.-J
-8 -7 -8 -5 -4 -3 -2 -1 0 2 3 4 5 8
mm

--+- 39,5 V
applied external voltage
1425 C, 354 glass. after 48 hours
800 mm2 electrode area
I!lIporim...,

162
168
.....
354
3,.
lime
bD.

48.0
48.0
"""ae
V

0.0
1,0
IlIrface
mm2

0
800
cation

N.W
NoW
.tartpt.
micro

1000
2000
C bulIt
Wpl%

6,81
6,47
C IOp/boI
WJlbI"

5,74
5"0
dis1aDCe
micro

700
18'0
eum.dilt.
micro

300
UO
Cloplbot
Wpl%

7,57
7,41
-
micro

600
300
eum.diJt.
micro

-300
-uo
CIOp/boI
wlbt"

2,49
3.14
-
micro

"00
~O
eum.dilt.
micro

-6000
-6000
-4000
ClOp/bot
wsb~

2,75
-
micro

2000

2000
cum.diat.
micro

-8000

-6000
143 3,. 49" 2" 800 N.W UOO 6,62 ',67 1200 300 8,20 4'0 -UO '.01 38'0 7-'0
140 3,. 48.0 5.0 800 N.20 1000 6,29 '''1 700 300 7,76 800 -~ 4-'0 4~ -'000 9,24 3000 -8000
127 354 48.0 39" 800 N.20 7SO 7,48 6,69 350 400 8-'0 550 -ISO 4,62 3850 -4000 13,99 4000 -8000

162 3" 48.0 0,0 0 NaW 1000 6,81 5,74 700 300 7"7 600 -300 2,49 "00 -6000 2,75 2000 -8000
144 3,. 48" 2" 20 N.20 4000 7,63 6"5 3950 SO 8,25 450 -400 3,57 S600 -6000
173 3,. 48.0 39" 20 N.W 7SO 6,29 5,40 350 400 7,93 550 -ISO 4,38 38SO -4000 6,73 2000 -6000

162 3,. 48.0 0.0 0 100 7SO 5,71 7;13 550 200 6.D9 700 -500 6,74 I~ -2000 1,25 6000 -8000
168 3,. 48,0 1,0 800 100 7SO 5,14 7,33 700 SO ',95 1'0 -100 7.00 1200 -1300 3,15 4700 -6000
143 3" 49" 2" 800 100 500 4,96 6,92 300 200 4,92 250 -SO 6,40 1450 -1500 2,61 4500 -6000
140 3,. 48.0 5.0 800 100 500 4,85 6,85 350 ISO 5.D6 300 -ISO 6,16 1850 -2000 2,32 6000 -8000
-8000
....
127 3,. 48.0 39" 800 100 7SO 4,78 6,93 600 ISO 5,43 300 -ISO ',99 3'0 -500 2.D6 7500 ell
U1
162 3,. 48.0 0.0 0 100 7SO 5,17 7;13 550 200 6,09 700 -500 6,74 1500 -2000 1,25 6000 -8000
144 3,. 48" 2" 20 100 200 5,13 6,83 400 -200 6;13 300 -500 7.00 800 -1300 4.D4 4700 -6000
173 3,. 48.0 39" 20 100 500 ',10 7,30 300 200 5-'0 350 -ISO 6,72 8SO -1000 2,13 '000 -6000

162 3,. 48.0 0.0 0 &0 1500 11,13 4,28 1300 200 5,45 3SO -ISO O.os 2OSO -2200 O.os S800 -8000
168 3" 48.0 1.0 800 &0 1'00 10,49 3,92 14SO SO 5,44 ISO -100 0,18 3900 -4000 0,18 2000 -6000
143 3,. 49" 2" 800 &0 1000 10,13 4,67 800 200 8,D1 2SO -SO om 2150 -2200 om 3800 -6000
140 3,. 48.0 5.0 800 &0 1500 10.00 4,34 1300 200 7;.2 350 -ISO 0.00 23SO -2500 0.00 5~ -8000
127 3,. 48.0
39" 800 &0 1000 10,48 4,18 800 200 6,10 350 -ISO 0,33 5850 -6000 0,64 2000 -8000

O.os ~
162 3,. 48.0 0.0 0 &0 1500 11,13 4,28 1300 200 M5 350 -uo O.os 2OSO -2200 S800 -8000
::l
144 3,. 48" 2-' 20 &0 300 9,61 300 0 4,74 300 -300 0,10 2700 -3000 0,10 3000 -6000 ::l
173 3" 48.0 39" 20 &0 1000 10,48 4.05 800 200 6,31 300 -100 0.D9 1900 -2000 0.D9 4000 -6000 (1)
><
162 3,. 48 0 0 A1203 1000 4,24 35,15 1500 -500 22"4 5500 -6000 23,24 2000 -8000 IV
168 3,. 48 I 800 AI203 500 6.00 34,68 700 -200 22,07 S800 -6000
143 3,. 49" 2" 800 AI203 750 6,37 35,47 8SO -100 25,95 3900 -4000 29,68 2000 -6000
140 3,. 48 5 800 AI203 7SO 6-'4 36,12 1050 -300 25.00 4700 -'000 32,43 3000 -8000
127 3" 48
39" 800 A1203 750 6,69 33.D2 900 -ISO 22,87 38SO -4000 36,48 4000 -8000
>-3
162 3,. 48 0 0 AI203 1000 4,24 35,15 1500 -500 22-'4 SSOO -6000 23,24 2000 -8000 III
144
. 173 ,,. 3$4 48"
48
2"
39"
20
20
AI203
AI203
200
500
5,65'
5,73
33,81
35.78
700
600
-500
-100
23,14
25"1
'500
3900
-6000
-4000 29,76 2000 -6000
lJ'
I-'
(1)

....
 Na20 Na20 Na20 N.20 Na20 Na20 Na20 Na20 Na20 Na20 Na20 Na20 Na20 Na20
EEpenment time temp. Slus parameter Itart pt. CODC. bulk CODC. top/bot dist. micro cum.disl. CODC. toplbot dill. lDicro cum.disl. CODC. top/bot disl. micro cum.disc. CODC. top/bot dial. micro cum.dill.

16 100 1350 354 750 8.00 6,65 600 ISO 8,65 ISO 0 4,04 3000 -3000
21 100 1425 354 750 7,00 6,60 400 350 8,00 550 -200 7.17 -2600
2600
26 100 !SOO 354 750 5.99 5,71 250 500 7,02 625 -125 6,72 1875 -2000
43 100 1350 354 Iher temp 750 7.90 6,62 600 ISO 8,60 300 -ISO 2,52 4850 -5000
28 100 1425 354 alter temp 1000 7,00 6,71 700 300 8,41 1300 -1000
32 100 !SOO 354 aller temp 400 5.94 5,94 0 400 8,06 3400 -3000 7,76 1000 -4000
SO 100 1350 354 w=O.2 750 7.87 6,49 550 200 8,89 500 -300 4,48 3400 -3700
25 100 1425 354 w=O,2 1000 7.00 6,34 700 300 8,42 350 -SO 7.29 3450 -3500
37 100 1500 354 w=O.2 0 6,00 6,00 0 0 8,25 3000 -3000
66 100 1350 354 w=2.0 400 7,68 6,74 200 200 8,65 400 -200 4,14 3300 -3500
62 100 1425 354 w=2.0 400 6,59 6,02 200 200 8,30 1000 -800 4,37 5200 -6000
67 100 1500 354 w=2.0 750 5,96 5,41 250 500 7,43 1000 -500 7,43 -4000
3500 6,68 1000 -5000
142
90
106
260
260
1350
1425
354'
354
2ncnewed
2ncocwed
SOD
SOD
7,94
6,99
7,D4
5,44
200
200
300
300
9.30
7.90
800
600
-500
-300 7,90 1700 -2000
1,88
6,30
7500
5000
-8000
-7000
...
0\
260 1425 354 2uenewed 750 6,88 6.09 650 100 8,14 2100 -2000 6,16 0\
3000 -5000
101 260 1425 354 2ucncwed SOD 7,21 6,40 350 150 8.03 250 -100 8,03 1900 -2000 7,03 3000 -5000
103 260 1425 354 2ucocwed 200 7.00 6,55 100 100 8,08 400 -300 8,08 1700 -2000 5,25 5000 -7000
133 600 1425 354' 6xrenewed 750 7,20 6,86 550 200 8,97 4200 -4000 8,46 1000 -5000
134 600 1425 354' 6xrencwed 0 6,81 6,81 0 0 8,38 1000 -1000 8,38 4000 -5000
81 100 1350 395 300 6,47 5,25 150 ISO 6.72 250 -100 1,83 3900 -4000
79 100 1425 395 400 5.60 5,05 200 200 6,78 1200 -1000 5,28 1000 -2000
80 100 1500 395 500 4.96 4,57 350 150 5.94 650 -500 5,07 4500 -5000
150 260 1350 395 2ucaewed 1000 6.70 5,39 1000 0 7,06 500 -500 1,30 ;l>'
8200 -8700 ::s
92 260 1425 395 2xrcncwed 500 5,49 4,95 350 150 6,15 300 -150 6,15 -2000
153 260 1500 395 2xrcncwed 750 4,61 4,44 350 400 6.63 4400 -4000
1850 3,25 6000 -8000 ::s
6.04 2000 -6000 (l)
>::
161 repeat 600 1350 395 6ucocwed 500 6.76 5.85 350 150 7,16 1150 -1000 4.18 5000 -6000 "l
161 600 1350 395 6ueaowed 750 5,53 4,56 600 150 5,17 450 -300 5.17 1700 -2000 2,72 4000 -6000
186 600 1350 395 6Deoewed 400 5,20 4.47 250 150 4,72 150 0 4,72 -2000
2000 3.63 4000 -6000
1.8 repeat 600 1425 395 6ucDewed 1000 4.99 4.67 1000 0 6,38 2000 -2000 5,74 4000 -6000
148 600 1425 395 6neDewed 750 5.83 4,58 900 -150 6.11 1850 -2000 6.11 -4000
2000 5,58 2000 -6000
187 260 1350 Litbhm 2ueaewed 750 2.39 2,39
>-3
0 750 4,24 4750 -4000 2.93 4000 -8000 III
180 260 1425 L1lhi1llll 2zrenewed 2000 2.11 2,11 0 2000 3,84 10000 -8000 tr
200 260 1500 Uthhu 2xrenewed 2000 3.39 2.93 500 1500 4,48 9500 -8000 ....
(l)

"l
 lC20 lC20 lC20 lC20 lC20 lC20 lC20 lC20 laO lC20 lC20 laO laO lC20
&perimenl time temp.
"ua panmeter Itart pt. cone. bulk CODe. toplbot dist. micro cum.disl. CODe. top/bot dill. micro cum.dill. coae. toplbot dill. micro cum.dial. CODe. toplbot dial. micro cum.dial.

16 100 13'0 3'4 500 ','3 7,'0 373 12' 6,00 300 -173 8,11 1&25 -2000 6,93 1000 -3000
21 100 1425 354 '00 4,9' 6,65 '7' -73 4,90 300 -373 7.11 2425 -2600
26 100 1500 3'4 750 4,77 ',4' 750 0 3,70 500 -500 4,39 1500 -2000

43 100 1350 3'4 IUerlomp 500 ',20 8,79 3'0 150 ',92 250 -100 7,64 1900 -2000 4,60 3000 -5000
28 100 1425 354 alter temp 400 4,92 6,54 300 100 4,32 300 -200 5,93 800 -1000
32 100 1300 '54 alter temp 600 4,77 5,76 500 100 3,76 600 -500 6,30 3500 -4000

50 100 13'0 3'4 .-0.2 500 ',14 7,'1 350 150 ',28 300 -150 7,20 18'0 -2000 6,26 1700 -'700
25 100 1425 354 w=O.2 '00 4.85 6,63 '50 1'0 4,19 300 -1'0 7.38 33'0 -3500
37 100 1500 354 .-0.2 500 4,67 5,55 4'0 50 4,03 1050 -1000 '.45 2000 -3000
66 100 1350 '54 .-2.0 400 4,92 7,23 3'0 '0 ',11 2'0 -200 6,97 1600 -2000 5,90 1300 -3'00
62 100 142' 354 w:;o;2.0 730 4,73 6,66 600 1'0 ',3' 350 -200 6,30 2800 -3000 4,90 3000 -6000
67 100 1300 '54 ,..-1.0 500 4,'6 ',76 350 150 3,65 400 -2'0 5.10 42'0 -4'00 4,90 '00 -'000

142 260 13'0 3'4· 2ucoowed 500 5,70 7,83 400 100 ',94 300 -200 8.90 2300 -2500 '.47 "00 -8000 ....
90 260 142' 334 2xreDewed '00 ',97 7,37 "0 -'0 5.90 650 -700 8,46 4300 -5000 8,33 2000 -7000 0'\
-..J
106 260 1425 3'4 :arencwed '00 ',86 7,13 600 -100 5,31 400 -'00 8,01 3'00 -4000 8,01 1000 -'000
101 260 1425 354 Zueoewcd 500 6,00 7,'9 450 '0 5," 350 -300 8,15 4700 -5000
103 260 1425 3'4 2uenewed 500 ',87 7,00 550 -'0 5,17 4'0 -500 8,13 3'00 -4000 8,13 3000 -7000

133 600 142' 3'4· 6uenewed 400 6,28 7,31 '00 -100 5,86 900 -1000 7,83 4000 -'000
134 600 142' 354· 6xrenewed 2000 6,20 7,41 2300 -300 5,89 700 -1000 7,26 4000 -5000

81 100 1350 395 300 7,29 11,32 200 100 9,39 300 -200 11.05 600 -1000 6,95 3000 -4000
79 100 1425 39' 400 6,90 10,10 2'0 150 7.34 300 -150 10,67 1&50 -2000
80 100 1500 39' 400 6,29 7,96 500 -100 6,32 900 -1000 8,70 3000 -4000 8,48 1000 -5000
:J:"
150 260 1350 395 Zuenewed 730 7,80 11,28 8'0 -100 10,86 400 -'00 11,30 2000 -2500 6.92 6200 -8700 ::s
92 260 1425 395 brenewed 500 8,22 10,90 550 -'0 8,09 250 -300 11,94 3700 -4000 10,73 4000 -8000 ::s
133 260 1'00 395 1uenewed 730 7,1& 8,82 1050 -300 6,32 700 -1000 10,30 5000 -6000 lD
><
161 "'pell 600 1350 39' 6uenewed 730 8,47 11,83 1250 -'00 -'00 11.83 3500 -4000 10.73 2000 -6000 I\J
161 600 1350 395 Urenewed 500 8,62 1&,54 800 -300 11,07 3700 -6000
1&6 600 1350 395 6uenewed 400 8,69 20,03 700 -300 12,01 3700 -6000
148 "'pell 600 1425 395 6uonewed 750 8,30 10,98 12'0 -500 9,47 '00 -1000 12,03 SOOO -6000
148 600 1425 395 6uenewed 750 8,11 10,73 9'0 -200 8,62 300 -500 12,12 "00 -6000
~
187 260 13'0 Ulhium 2zrenewed 1000 3,17 3,81 1200 -200 2,81 600 -1000 ',17 5000 -6000 4,95 2000 -8000 III
180 260 142' Ulbium 2uenewed 4000 3,17 3,66 3700 300 2,30 1600 -1300 4,87 6700 -8000 t:r
I-'
200 260 1500 Ulhium :arenewed 200 3,10 3,10 0 200 2,13 4200 -4000 3,13 4000 -8000 lD
....
 B,O B,O B,O B.O B,O B,O B.O B.O B,O B.O B.O B,O B,O B.O
Experiment time temp. alul parameter ,tart pt. COIle. bulk CODe. toplbot dist. micro cum.dist. CODe. toplbot dist. micro cum.diat. CODC. toplbot dist. micro cum.dist. CODe. toplbol· diat. micro· cum.disl.

16 100 1350 35. 750 lo.a3 U8 600 150 6,90 250 -100 M5 1900 -2000 0 1900 -2000
21 100 1.25 35. 750 9,82 6,10 850 -100 7,U 200 -300 0,71 2500 -2800 0 2500 -2800
26 100 1500 35. "0 9,00 6,51 750 0 11,38 1000 -1000 7,81 500 -1500

.3 100 1350 35. aUer temp 7S0 10,7. S,OO 600 150 6,60 300 -150 0,15 2850 -3000 2000 -2150
28 100 H25 35. alter temp 62S 10,05 5,82 SOO 125 9,2S 300 -17S U4 825 -1000
32 100 lSoo 35. alter temp 7S0 9,07 S,87 600 150 9,97 650 -500 0,7. 3500 -.000 3500 -4000

50 100 1350 3S. ",-0.2 750 10,97 4,66 600 150 7,97 300 -150 0,30 35S0 -3700 0 23S0 -2S00
2S 100 1425 354 ",.0.2 6S0 10,21 5,5S SOO ISO 10,25 300 -ISO O,S5 3350 -3S00 0 33S0 -3S00
37 100 lS00 3S. 91=0.2 600 9,01 6,30 .SO ISO 9,53 1150 -1000 2,99 2000 -3000 0 3000 -4000
66 100 1350 35. w=2.0 500 11,13 4,89 4S0 50 8,19 250 -200 0,35 3300 -3SOO 0 2000 -2200
62 100 142S 35. 91'-2.0 750 10,18 S,29 S50 200 M6 .00 -200 O,2S S800 -6000 0 3000 -3200
67 100 1500 3S. w a 2.0 500 9,19 6,31 350 150 10,67 .SO -300 0,28 4700 -5000 0 4000 -4300

142 260 1350 3S.· brenewed f-'

90 260 142S 3S. brenewed 500 9,90 S,84 4S0 50 8,21 5S0 -SOO 0,07 6S00 -7000 0 4000 -.SOO 0'1
106 260 1.25 35. Zuenewed 750 10,11 S,25 850 -100 8,35 400 -SOO 0,11 .SOO -5000 0 3000 -3S00 CXl
101 260 142S 3S4 Zarencwed 7S0 10,27 5,51 6S0 100 9,08 .00 -300 O,6S 4700 -SOOO 0 .700 -5000
103 260 1425 3S4 2Erenewed 750 9,99 6,25 700 SO 9,61 550 -SOO 0,25 6S00 -7000 0 .500 -5000

133 600 1.2S 3S.· 6uenewed

134 600 1.25 3S.· 6zrenewed

81 100 1350 395 7S0 8,6S 2,85 6S0 100 3,81 300 -200 0,20 3800 -.000 0 1200 -1400
79 100 142S 39S SOO 8,S1 .,01 350 150 6,23 300 -ISO 0,76 18S0 -2000 0 1500 -1650
80 100 1500 39S 400 7,03 4,5S SOO -100 S,75 .00 -SOO 0,19 SSOO -6000 0 3S00 -4000
~
150 260 1350 39S 2D'eaewed 1000 8," 2,77 1100 -100 3,30 200 -300 0,07 3700 -4000 2000 -2300 ::l
92 260 142S 39S 2uenewed SOO 7,66 3,82 SOO 0 6,2S 300 -300 0,17 S700 -6000 4000 -4300 ::l
(1)
lS3 260 1500 39S 2Deoc'*ed 1000 7," 4.60 1150 -150 6,86 8S0 -1000 O,S8 SOOO -6000 SOOO -6000
><
181 r.pelt 600 1350 395 6xrenewed 750 8,19 2,88 8S0 -100 3,51 100 -200 0,27 3800 -4000 0 2000 -2200 l\J
161 600 1350 395 menewed 7S0 8,30 0,42 10S0 -300 1,18 700 -1000 0,04 SOOO -6000 0 1500 -2SOO
186 600 1350 39S 6ueaewed SOO 7,77 3,22 SSO -SO 3,90 100 -150 0,08 5850 -6000 0 2000 -21S0
148 r.pelt 600 142S 395 6xrenewed 7S0 7,64 3,60 1050 -300 5,05 700 -1000 0,54 SOOO -6000 0 4S00 -SSOO
148 600 142S 395 6uenewed 7S0 7,37 3,06 900 -ISO S,44 3S0 -SOO 0,00 SSOO -6000 0 4000 -4500
1-3
187 260 13S0 U.hhm 2xren.w.d lSoo 7,86 6,74 1700 -200 8,04 800 -1000 O,OS 7000 -8000 0 3SOO -4S00 l»
180 3S00 7,94 -8000
t:r
260 142S U'hhm 2....n.w.d 4000 8,02 6,12 SOO 1000 -SOO 0,44 7S00 0 6SOO -7000 f-'
200 280 1500 Uthhau 2u'eDewed 4000 6,57 6,11 3500 500 7,43 1500 -1000 2,85 7000 -8000 0 11800 -12800 (l)

~
 srO srO SrO srO SrO Sro SrO SrO Sro Sro srO SrO Sro srO
Experiment time temp...... parameter ,tart pt. oooc. bulk CODC. topfbot dis•. mIcro cum.disl. conc. toplbot disl. micro cum.dist. CODe. toplbot dial. micro cum.disl. CODe. toplbot· dbt. miao. cum.disl.

16 100 1350 354 500 3,73 1,22 400 100 1,76 200 -100 0,18 1900 -2000 0 1900 -2000
21 100 1425 354 500 3,27 1,53 350 150 1,76 150 0 0,15 2800 -2800 0 3000 -3000
26 100 1500 354 1000 3,45 1,85 1000 0 2,99 1000 -1000 1,85 500 -1500
43 100 1350 354 alter temp 750 3,57 1,20 650 100 1,40 200 -100 0,00 2900 -3000 1900 -2000
28 100 1425 354 liter temp 500 3,46 1,46 450 50 2,00 200 -150 0,78 850 -1000
32 100 1500 354 Iltertemp 1000 3,14 1,16 1000 0 2,50 500 -500 0,60 3500 -4000 0 3500 -4000
50 100 1350 354 "'-0.2 750 3,81 1,10 700 50 1,50 100 -50 0,50 -2000
1950 0 2500 -2550
25 100 1425 354 "'-0.2 600 3,57 1,43 500 100 2,50 300 -ZOO 0,31 -3500
3300 0 3300 -3500
37 100 1500 354 .-0.2 400 3,02 1,80 500 -100 2,50 900 -1000 0,82 2000 -3000 0 3000 -4000
66 100 1350 354 .-2.0 500 3,76 1,27 500 0 1,37 500 -500 0,04 3000 -3500 0 2000 -2500
62 100 1425 354 w=2.0 1000 3,38 1,28 1000 0 2,40 300 -300 0,00 5700 -6000 0 3000 -3300
67 100 1500 354 _=2.0 1500 3,10 1,93 1500 0 Z,78 300 -300 0,06 4700 -5000 0 4000 -4300
142 260 1350 354· 2xrenewed
I-'
90 260 1425 354 2xreoewed 500 3,31 1,43 700 -200 Z,IZ 300 -500 0,00 5500 -6000 /J\
0 3000 -3500
106 260 1425 354 2xrenewed 500 3,38 1,23
101
103
260
260
1425
1425
354
354
2xrenewed
2xrenewed
750
500
3,38
3,30
1,38
1,50
700
750
550
-200
0
-50
1,95
1,82
100
300
-300
-300
0,00
O,ZO
4700
4700
-5000
-5000
0
0
2700
4500
-3000
-4800 '"
1,97 450 -500 0,06 4500 -5000 0 4000 -4500
133 600 1425 354· 6xronewed
134 600 1425 354· 6xrenowed

81 100 1350 395 400 11,26 2,54 350 50 2,54 250 -200 0,06 3800 -4000 0 800 -1000
79 100 1425 395 750 11,41 4,24 550 ZOO 4,24 700 -500 0,79 1500 -2000 0 1500 -2000
80 100 1500 395 400 9,24 3,76 600 -200 5,59 300 -500 0,51 5500 -6000 0 3500 -4000
150 260 1350 395 2xrenowed 1000 11,57 1,92 1100 -100 2,07 100 -ZOO 0,00 600 -800 0 600 -800 ~
92 260 1425 395 2ueaewed 500 9,74 3,31 550 -50 5,33
::l
450 -500 0,41 3500 -4000 0 2500 -3000 ::l
153 260 1500 395 2xreoewed 1000 10,55 4,37 1150 -150 7,57 850 -1000 0,04 5000 -6000 CD
0 3500 -4500
>C
161 repelt 600 1350 395 6xrenewed 750 9,05 1,35 850 -100 1,61 100 -200 0,00 1800 -2000 0 1200 -1400
161
186
600
600
1350
1350
395
395
6xrenewed
6:uenewed
750
500
11,74
11,01
0,08
1,97
1050
600
-300
-100
0,72
2,64
700
50
-1000
-150
0,00
0,00
3000
5850
-4000
-6000
0
0
1000
1200
-2000
-1350
'"
148 repelt 600 1425 395 6xrenewed 1000 10,74 3,04 1300 -300 4,21 700 -1000 0,00 3000 -6000 0 4000 -5000
148 600 1425 395 6xrenewed 750 10,05 1,96 950 -200 4,59 300 -500 0,00 5500 -6000 0 2000 -2500
187 260 1350 Lithinm 2xrenowed
t-3
1000 9,45 5,40 1200 -200 6,73 800 -1000 OM 3000 -4000 0 3000 -4000 III
180 260 1425 Lithium 2xreaowed 4000 9,82 5,76 3500 500 8,00 2000 -1500 0,08 6500 -8000 0 5000 -6500 tr
200 260 1500 Lithium 2xrenewed 0 5,86 5,86 0 t-'
0 6,64 2000 -2000 2,77 6000 -8000 0 10300 -12300 CD
U1
Conceotration profiles o[ A1203.

E~rimeDt time temp. ,I... parameter .tan pt. bulk cone. dist. micro cum.disl. CODC. toplbot dial. micro cum.. dill. CODC. toplbot disi. micro cum. dial CODC. toplbot

16 100 1350 '54 750 5,57 850 -100 '5.91 2900 -'000 22.87
21 100 1425 '54 500 8.'8 900 -400 36.60 2400 -2800 31.15
26 100 1500 )54 500 11.70 900 -400 '5.50 1600 -2000 35.50 I-'
-..J
50 1350 3~" w. 0.2 500 750 -250 36.95 -)700 0
100 5,2' '450 24.41
25 100 1425 3~4 w • 0.2 600 7.08 1100 -500 '8.49 '000 -3500 '1.77
'7 100 1500 '54 " - 0.2 500 10.94 1500 -1000 '6.07 2000 -3000 '6.07

66 100 1350 '54 w _ 2.0 500 5.40 700 -200 36.55 "00 -'500 23.15
62 100 1425 '54 " - 2.0 750 7.45 950 -200 36.95 5800 -6000 26.47
67 100 1500 '54" - 2.0 450 10,20 750 -300 '6.77 4700 -5000 '0.84

81 100 1350 '95 400 2.'1 600 -200 35.32 '800 -4000 22.08
79 100 1425 '95 500 '.34 650 -150 )7.14 1850 -2000 '1.12
80 100 \SOO '95 400 8.84 1400 -1000 37,56 4000 -5000 '1.08 ~
::s
150 260 1350 395 1000 1.'8 1500 -500 '4.17 8200 -8700 18.09 ::s
92 260 1425 )95 500 5,50 800 -300 38,42 7700 -8000 25.66 CD
153 260 1500 )95 750 6.97 1750 -1000 39.19 5000 -6000 35.74 X
!\J
161 600 H50 395 750 1.61 1050 -300 35.93 5700 -6000 25.'0
186 600 1350 '95 750 ),47 1050 -'00 35.87 5700 -6000 28.87
148 600 1425 )95 750 4.64 1750 -1000 39.75 5000 -6000 '6.07

186 600 H50 )95 bottom )00 2,08 450 -150 '5,42 850 -1000 '5.42 7000 -8000 22.11 t-3
III
tl'
.....
CD
0'1
 171

Annex 3

Table of calculated e.e.h. of the crucible experiments.

An example of the calculation of the e.e.h. (electron equivalents

per hour) :
As example the calculation of the e.e.h. of potassium in the
experiment 161 (first row of the table) will be given.
In the glass (cullet) before the start of the experiment, the
content of K20 is 9.09 weight%, after the glass melt was 90.5
hours in contact with the fused cast AZS crucible at 1350 °e, the
K20 content in the glass was 7.80 weight% (average of a duplicate
measurement). The difference between the K20 content before and
after the glass melt contact with the AZS crucible is 1.29
weight%.
The average mole weight of the glass is 69.92 and the mole weight
of K20 is 94. The change in K20 in the glass is therefore
(1.29*69.92}/94 = 0.96 mole%.
The contact time is 90.5 hours, the decrease of K20 per hour is
0.96/90.5 = 10.6 10~ mole%/hour.
The decrease in K20 in the glass melt represents the diffusion of
two K+ cations in the glassy phase of the AZS and, to keep the
electro neutrality conditions, two electrons. So the change of the
K20 content in the glass melt in the crucible gives a migration of
2*10.6 10~ = 21.2 10~ electrons/hour.
The measured values are only valid for the standard test set-up
and can only be used as a relative value, it is therefore more
convenient to multiply all so calculated values with 10+ 3 and use
these values as electron equivalents per hour (e.e.h.).
The e.e.h. of K20 in the experiment 161 is therefore 21.2 e.e.h.
 172

Annex 3 Table 1

ezperimenl glass method temp della time aver. time li20 el.eq/br Na20 el.eq/hr K20 el.eq/hr BaO el.eq/hr SrO el.eq/hr

161 395 eulIet 1350 90,50 45,250 -4,24 21,20 2,32 1,6
161 395 cuUet 1350 72,50 126,750 0,00 16,00 9,33 11,5
161 395 cullet 1350 95,50 210,750 2,24 12,16 7,37 12,8
161 395 collet 1350 72,50 294,750 4,20 9,54 7,12 10,2
161 395 cuUet 1350 95,50 378,750 3,30 6,94 4,40 9,0
161 395 collet 1350 72,00 462,500 5,02 7,86 3,11 6,6
186 395 collet 1350 92,00 46,000 -5,88 28,62 5,76 8,2
186 395 cuUet 1350 69,75 126,/175 -5,16 23,78 8,45 17,3
186 395 collet 1350 97,75 210,625 -1,84 13,46 4,35 7,6
186 395 collet 1350 72,75 295,/175 4,20 12,06 7,29 13,4
186 395 collet 1350 95,50 380,000 2,96 10,04 3,45 7,3
186 395 cullet 1350 72,00 463,750 2,02 9,20 1,46 3,7
186 39' powder 1350 92,00 46,000 -7,96 36,78 0,05 3,3
186 395 powder 1350 69,75 126,/l75 -2,26 32,64 0,92 2,0
186 395 powder 1350 97,75 210,625 0,82 19,40 0,47 1,1
186 395 powder 1350 72,7' 295,/175 9,30 14,42 3,64 8,5
186 395 powder 1350 95,50 380,000 6,86 9,58 2,06 6,2
186 395 powder 1350 72,00 463,750 8,46 10,22 2,28 4,3
150 395 collet 1350 92,00 46,000 -4,16 25,96 1,79 1,1
150 395 cuUet 1350 70,20 127,100 -4,66 24,90 2,54 4,6
148 395 cullel 1425 88,25 44,125 4,08 30,52 4,76 8,5
148 395 collet 1425 72,25 124,375
148 395 collet 1425 95,75 208,375 11,78 18,48 4,68 8,4
148 395 collet 1425 72,25 292,375 14,98 20,18 4,55 7,0
148 395 collet 1425 95,75 376,375 14,36 14,30 3,15 5,3
148 395 collet 1425 72,25 460,375 14,68 14,42 4,43 6,8
148 395 powder 1425 88,25 44,125 5,76 35,82 6,01 11,0
148 395 powder 1425 72,25 124,375
148 395 powder 1425 95,75 208,375 11,30 20,28 5,15 10,1
148 395 powder 1425 72,25 292,375 12,64 21,00 5,00 9,7
148 395 powder 1425 95,75 376,375 12,72 15,00 4,73 8,7
148 395 powder 1425 72,25 460,375 14,98 16,06 5,25 9,7
85/92 395 cuUet 1425 90,00 45,000 8,60 33,80 8,40 13,4
85/92 39' collet 1425 70,00 125,000 4,60 29,40 5,10 9,8
174 395 collet 1500 88,00 44,000 35,50 45,56 8,83 15,6
174 395 collet 1500 71,50 123,750 33,28 40,06 8,82 15,0
174 395 collet 1500 95,50 207,250 26,46 25,16 7,42 14,3
174 395 collet 1500 73,00 291,500 26,26 24,92 7,73 13,5
174 395 collet 1500 96,00 376,000 18,20 16,54 6,62 11,1
174 395 collet 1500 71,50 459,750 21,76 21,36 6,78 12,1
174 395 powder 1500 88,00 44,000 35,38 53,50 10,18 17,9
174 395 powder 1500 71,50 123,750 31,08 46,72 9,27 16,2
174 395 powder 1500 95,50 207,250 24,32 28,36 8,04 14,8
174 395 powder 1500 73,00 291,500 15,14 26,70 8,70 16,0
174 395 powder 1500 96,00 376,000 11,98 19,44 8,66 16,6
174 395 powder 1500 71,50 459,750 16,24 26,84 6,26 8,4
201 395 powder 1500 89,00 44,soo 33,36 55,24 10,27 18,8
201 395 powder 1500 72,00 125,000 35,40 51,24 10,54 18,9
201 395 powder 1500 94,25 208,125 27,28 30,70 8,58 16,1
201 395 powder 1500 72,25 291,375 24,04 30,48 7,59 12,8
153 395 collet 1500 88,08 44,042 30,22 42,06 8,46 13,4
153 395 collet 1500 72,83 124,500 34,38 40,54 9,91 17,0
 173

Annex 3 Table 2

ClperimcDt g1aa mcthod tcmp dclta timc aver. time Li20 clccq/hr Na20 clccq/hr K20 clccq/hr B.O clccq/hr srO clccqlh

142 3S4 cullct 1350 90,00 45,000 6,00 19,&l

142 3S4 cullct 1350 72,00 126,000 2,80 16,60
109 3S4 cullct 1425 90,00 45,000 22,&l 29,40 5,20 0,3
109 3S4 cullct 1425 71,00 125,500 18,60 26,00 3,90 1,1
109 3S4 cullct 1425 95,00 208,500 13,&l 17,00 4,30 1,7
109 354 cullct 1425 71,00 291,500 13,60 16,00 4,00 1,8
109 354 cullct 1425 97,00 375,500 11,40 9,60 3,00 1,8
109 354 cullct 1425 71,00 459,500 9,00 8,40 2,80 0,9
133 354 cullct 1425 89,50 44,750 17,&l 25,40
133 3S4 cullct 1425 71,00 125,000 18,&l 25,40
133 354 cullct 1425 97,00 209,000 13,60 14,&l
133 3S4 cullct 1425 72,00 293,500 12,&l 16,00
133 3S4 cullet 1425 96,00 377,500 13,40 9,60
133 354 cullet 1425 75,00 463,000 8,20 8,60
134 3S4 cullct 1425 89,25 44,630 19,00 28,00
134 354 cullct 1425 72,00 125,250 25,00 26,&l
134 354 cullct 1425 96,25 209,380 13,80 15,60
134 354 cullct 1425 71,75 293,380 14,40 15,20
134 354 cullct 1425 96,00 377,250 12,60 10,00
134 354 cullet 1425 70,25 460,380 13,80 10,80
102 3S4 cullct 1425 90,00 45,000 23,00 25,60 14,30 6,6
102 354 cullct 1425 71,00 125,500 22,20 26,&l 4,80 3,5
104 354 cullct 1425 90,00 45,000 18,00 26,&l 6,20 3,3
104 354 cullct 1425 72,00 126,000 17,60 23,60 6,10 3,6
106 354 cullct 1425 89,00 44,500 20,60 28,40 9,20 5,4
106 3S4 cullct 1425 73,00 125,500 19,00 25,20 5,90 3,9
108 354 cullct 1425 89,00 44,500 19,&l 24,00 16,30 5,6
108 354 cullct 1425 71,00 124,500 17,20 25,40 3,20 1,3
112 3S4 cullct 1425 89,00 44,soo 19,&l 27,00 8,30 3,2
112 3S4 cullct 1425 71,00 124,500 15,40 23,00 4,10 1,5
101 354 cullct 1425 90,00 45,000 19,20 27,80 6,30 4,1
101 354 cullct 1425 72,00 126,000 18,00 25,60 2,20 2,5
103 354 cullct 1425 90,00 45,000 18,20 27,&l 3,30 1,9
103 354 cullct 1425 72,00 126,000 21,00 26,60 5,50 3,5
105 354 cullct 1425 88,00 44,000 17,20 27,20 5,70 3,6
105 354 cullet 1425 74,00 125,000 17,40 23,80 4,70 3,4
139 3S4 cullct 1425 22,50 11,250 14,60 59,00
139 3S4 cullct 1425 24,00 34,500 16,00 30,40
139 3S4 cullct 1425 24,00 58,500 10,00 27,60
187 lithium powder 1350 93,03 46,515 34,34 -3,06 15,56 6,20 9,3
187 lithium powdcr 1350 70,39 128,224 30,16 3,44 13,10 8,26 13,0
180 lithium powder 1425 90,75 45,375 57,86 1,72 18,84 11,93 18,6
180 lithium powdcr 1425 70,42 125,959 45,36 10,02 19,14 10,67 15,4
200 lithium powder 1500 88,83 44,417 85,76 13,16 21,88 19,15 32,0
200 lithium powdcr 1500 71,58 124,625 68,36 22,20 21,50 18,84 23,9
 174

Annex 4

Bubbles analysis.

In chapter 5 already is explained that the bubbles are analyzed by

breaking a bubble in a high vacuum chamber and analyze the
chemical gas composition of the bubble with a Balzers QMG 420
quadrupole mass spectrometer. The diameter of the analyzed bubbles
is calculated using the volume of the bubbles. The volume of the
bubble is calculated using the increase in the summation of the
partial pressures of the known volume of the high vacuum chamber.
The diameter of the prepared bubbles are measured by a microscope.
The results of the measurements are given in the table of this
annex.

Explanation of the table.

Exper. no. number of the test.

Time hrs. test duration in hours.
Temp. degree C temperature in 0 C.
Glass used glass (cullet).
Bubbles no. counted number of bubbles in the sample.
Volume cm3 sample volume in which the bubbles are
counted.
Bubbles N/cm3 counted number of bubbles per cm3 •
Range mm the arithmetical average diameter in rom of
the five largest bubbles found in the sample.
Number prep. number of bubbles prepared for analysis.
Diameter mm arithmetical average diameter in rom of the
prepared bubbles.
Number anal. number of bubbles successfully analyzed.
Diameter mm arithmetical average diameter in mm of the
analyzed bubbles.
the arithmetical average gas composition
of the analyzed bubbles.
Elper. TIme Temp. Glass Bubbles Volume Bubbles Range Number Diameter Number Diameter 02 N2 CO2 AI
no. bu. degree C no. em3 N/em3 mm prep. mm anal. mm
16 100 1350 354 161 7,42 21,70 0,332 5 0,184 2 0,180 0,0 82,6 17,4
21 100 1425 354 338 15,52 21,78 0,445 36 0,238 7 0,229 71,6 28,1 0,0 0,3
26 100 1500 354 309 11,97 25,81 0,403 54 0,203 24 0,215 93,5 5,1 1,5 0,0
43 100 1325-1375 354 32 5,91 5,41 0,229 30 0,121 2 0,160 0,0 89,5 10,0 0,6
28 100 1400-1450 354 212 6,32 33,54 0,383 41 0,186 13 0,217 46,2 7,1 46,6 0,1
32 100 1475-1525 354 78 13,09 5,96 0,621 31 0,221 9 0,267 99,0 1,0 0,0 0,0
78 100 1350 354 48 6,72 7,14 0,331 31 0,113 12 0,120 0,0 81,5 17,7 0,8
71 100 1425 354 118 7,40 15,95 0,396 38 0,128 14 0,169 50,1 47,4 2,1 0,5
69 100 1500 332 9,61 34,55 0,611 37 0,273 27 0,289 78,6 20,3 1,0 0,1
81 100 1350 395 30 10,41 2,88 0,232 42 0,122 8 0,120 0,3 78,0 20,9 0,8
79 100 1425 395 85 9,99 8,51 0,381 33 0,116 10 0,099 10,1 53,6 35,7 0,5
80 100 1500 395 389 9,73 39,98 0,497 34 0,196 18 0,204 65,7 30,1 4,0 0,3
93 100 1350 354ZnO 43 11,52 3,73 0,322 32 0,113 18 0,109 0,0 90,1 9,2 0,7
89 100 1425 354ZnO 157 10,95 14,34 0,540 40 0,145 19 0,179 31,2 60,9 7,5 0,5
94 100 1500 354ZnO 568 9,68 58,68 0,602 32 22 0,217 SO,5 18,6 0,8 0,1 I-'
90 260 1425 354 -..J
320 11,39 28,09 0,380 32 0,195 17 0,220 75,4 17,2 7,2 0,2 (J1
106 260 1425 354 513 12,52 40,97 0,432 38 0,179 17 0,213 70,0 27,3 2,3 0,3
92 260 1425 395 165 10,24 16,11 0,430 32 0,232 10 0,306 79,4 18,7 1,8 0,1
102 260 1425 354395 483 10,35 46,67 0,465 36 0,134 18 0,163 24,2 65,7 9,5 0,6
104 260 1425 35439SNa/K 345 11,36 30,37 0,409 30 0,193 12 0,235 58,9 39,2 1,6 0,4
128 260 1425 35439m 321 12,24 26,23 0,420 31 0,193 16 0,217 62,6 35,4 1,7 0,4
117 260 1425 35439SZr 231 13,00 17,77 0,442 41 0,127 24 0,150 40,2 53,8 5,5 0,6
112 260 1425 354395mu 322 11,08 29,06 0,427 35 0,214 21 0,233 67,6 30,1 2,1 0,2
125 260 1425 354354Zn 230 10,77 21,36 0,503 32 0,187 15 0,158 59,4 39,1 0,9 0,4
108 260 1425 354TA3779 414 11,92 34,73 0,434 37 0,174 15 0,188 58,8 36,6 4,2 0,4
142 260 1350 354 425 25,11 16,93
150 260 1350 395 130 11,06
0,343
0,251
27
24 0,083
15
6
0,188
0,097
4,3
0,0
21,7 73,4 0,2 >
::l
31,6 68,0 0,3
153 260 1500 395 423 13,23 31,97 0,467 36 0,271 30 0,239 88,8 10,4 0,7 0,1 ::l
187 260 1350 CD
Lilhium 290 14,27 20,32 0,507 32 0,233 26 0,242
180 260 1425 Lilhium
66,7 30,6 2,4 0,3 ><
58 28,22 2,06 0,601 28 0,261 23 0,274 76,8 22,6 0,4 0,2
200 260 1500 Lilhium 27 01>
20,59 1,31 0,657 21 0,401 16 0,413 91,0 8,5 0,4 0,1
133 600 1425 354 221 11,01 20,07 0,450 29 0,182 17 0,206 92,4 6,6 0,6 0,0
161 600 1350 395 261 11,30 23,10 0,221 30 0,140 15 0,148 0,1 59,9 42,5 0,4
186 600 1350 395 75 13,70 5,47 0,189 18 0,114 9 0,140 24,8 43,0 31,8 0,4
148 600 1425 395 379 14,58 25,99 0,376 24 0,163 6 0,187 79,0 18,7 2,1 0,2 8
149 100 1350 426 393 12,08 32,53 0,382 32 0,176 21 0,199 49,3 38,6 CI
11,7 0,4 tr
147
146
100
100
1425
1500
426
426
696
622
11,71
12,50
59,44
49,76
0,461
0,566
32 0,219' 24 0,216 88,1 9,8 2,0 0,1 ....
CD
32 0,263 21 0,227 93,5 6,1 0,4 0,1
214 100 1350 433 113 9,52 11,87 0,181 32 0,103 14 0,112 0,0 86,5 12,7 0,8 I-'
213 100 1425 433 439 8,95 49,05 0,432 32 0,140 23 0,141 56,5 42,3 0,8 0,4
212 100 1500 433 612 9,33 65,59 0,463 37 30 0,210 86,8 12,7 0,4 0,1
Glass composition and properties.
For definitions see annex 1.
Glass

composition 395NaIK 395Ti 395Zr 395mu 354Zn TA1779 426 433

Si02 61,78 62,77 64,55 58,75 63,45 59,27 63,80 59,60 weight%
A1203 0,76 0,75 0,88 4,58 3,36 3,36 3,30 3,20 weight%
Li20
0,20 weight%
Na20 9,34 6,75 6,81 6,81 9,44 9,23 7,30 7,80 weight%
K20 7,03 9,19 9,07 8,87 6,18 6,54 7,30 7,70 weight%
MgO 0,28 0,29 0,25 0,30 0,98 0,25 0,40 weight%
CaO 0,41 0,40 0,40 0,38 1,40 1,51 0,35 0,60 weight%
SrO 9,38 0,94 9,23 9,29 2,95 8,91 7,00 8,60 weight%
BaO 7,84 7,78 7,78 7,87 10,51 8,38 8,90 9,60 weight%
Zr02 2,09 2,06 2,07 2,07 2,50 weight%
Ti02 0,42 0,42 0,43 weight%
Sb203 0,38 0,37 0,38 0,37 0,37 0,38 0,40 0,40 weight%
properties I-'
-..J
Density 2752 2,731 2,709 2,764 2,782 2,70 2,77 kg1m3 0\

Viscosity:
strain point 487 482 474 499 501 465 495 C
annealing pom 521 517 504 533 532 505 530C
softening point 704 706 675 717 703 685 715C
working point 1009 1026 1002 1045 1012 1025 1040 C
melting point 1416 1447 1430 1470 1418 1485 1460 C
Constants
Fulcher
equation:
A -1,874 -2,036 -2,021
B 4874,471 5142,712 4917,760
C 172,248 196,079 195,142
>
~
~
CD
X
• Calculated density and viscosity U1
 177

JI.nnex 6

student or t-test.

For testing the difference of two means of small samples, the

Student or t-test may be used.
~~he theory of Student's t distribution requires one to assume that
t:he basic variables Xl and x 2 possess independent normal
distributions with equal standard deviations. If these assumptions
are reasonably satisfied then one may treat the variable t as a
Student t variable with nl + n 2 - 2 degrees of freedom.
In the case of the evaluation of the bubble analysis of the
crucible experiments the means of the different groups of bubble
analysis are tested assuming that there is no difference in the
Dleans. In all the tests a significance level of 0.05 has been
chosen.
,]~he used calculation method is describe below.

i.n which
XI mean of distribution 1.
X2 mean of distribution 2.
C'
~'I standard deviation of distribution 1.
C'
~'2 standard deviation of distribution 2.
NI number of values of distribution 1.
N2 number of values of distribution 2.

Literature reference: P.G. Hoel, Elementary statistics,

Fourth edition 1962,
J. Wiley & Sons, New York.
samenvatting

De interactie van het gesmolten glas met het vuurvast materiaal

van de industriele glassmeltwannen heeft een grote invloed op de
kwaliteit van het geproduceerde glas. Het vuurvast materiaal dat
het meest gebruikt wordt in de glasindustrie, in direct contact
met gesmolten glas, is het smeltgegoten AZS (Aluminiumoxyde
Zirconiumoxyde silicaat). Als gevolg van de interactie van het
gesmolten glas met het smeltgegoten AZS kunnen verschillende
soorten inhomogeniteiten zoals stenen, knobbels en bellen in het
glas gevormd worden.
Het doel van deze studie is het vinden, voor het geval van
smeltgegoten AZS in contact met gesmolten glas, van het
mechanisme van knobbel- en bel-vorming en de parameters die dit
mechanisme beYnvloeden.
Het uiteindelijke doel is de ontwikkeling van een (semi)
kwantitatief model dat de hoeveelheid glasfouten voorspelt ten
gevolge van het contact van het gesmolten glas met het
smeltgegoten AZS, zodat men in staat is de omstandigheden te
kiezen met de laagste vormingssnelheden van deze glasfouten.
De meeste bellen die ontstaan bij temperaturen boven 1400 ·C zijn
oorspronkelijk zuurstof bellen, veroorzaakt door een
electrochemisch mechanisme in het vuurvast materiaal. Wanneer
gesmolten glas in contact gebracht wordt met smeltgegoten AZS,
transporteert de diffusie van kationen van de glassmelt in het
AZS positieve lading in het inwendig van het AZS. De diffusie van
de kationen wordt veroorzaakt door de lagere partiele Gibbs
energie van de glassmelt kationen in het hoog Al 20 3 houdende
glasfase van het AZS vuurvast. De positieve lading ten gevolge
van de kation diffusie wordt in balans gehouden door elektronen
die van de glassmelt/AZS overgang naar het inwendige van het AZS
bewegen om electroneutraliteit te handhaven. De elektronen
afkomstig van de zuurstofionen in de glassmelt reduceren de
polyvalente ion onzuiverheden (voornamelijk ijzer) van de
smeltgegoten AZS. De snelheid van zuurstofgas vorming bij de
vuurvast/glassmelt overgang, als een resultaat van de
redoxreactie (elektronen donor), wordt bepaald door de diffusie
!;nelheid van de kationen van het gesmolten glas in de glasfase
(de bindende fase) van het smeltgegoten AZS.
De zuurstofbellen ontstaan in de overgangslaag tussen
!;meltgegoten AZS en de glassmelt. De overgang bestaat uit een
~Jlasachtige laag van produkten, die resulteren uit de reacties
1~ussen smeltgegoten AZS en de glassmelt, met nodulair Zr02 aan de

,rourvast zijde. Een bel ontstaan in deze overgangslaag drukt zich

door deze laag in de glassmelt, daarbij wordt soms ook een
~Jedeelte van de AI 20) rijke laag in de bulk van de glassmelt

~Jedrukt.

net hoge AI 20) gehalte van de glasachtige gedeelte van de

()vergangslaag geeft deze "deeltjes" een zo hoge oppervlakte-
spanning dat ze bolvormige glasachtige knobbels in de glassmelt
vormen.
Dit betekent dat een enkele mechanisme verantwoordelijk is voor
de vorming van zowel bellen als knobbels aan de overgang van
~Jlassmelt en smeltgegoten AZS.

net snelheid bepalende element van dit proces, de kation

diffusie, is gemeten voor de belangrijkste oxyden van een
hedendaags TV-scherm glas (Li 20, Na20, K20, BaO, SrO, CaO en MgO) •
net effect van Li 20 op de zuurstofbel vorming is erg groot, niet
aIleen zijn eigen bijdrage, maar ook veroorzaakt Li 20 dat ander
clxyden een groter bijdrage hebben dan dat deze oxyden zouden
hebben in een glas zonder Li 20. Veruit de grootste bijdrage van
de andere oxyden, behalve Li 20, heeft K20 vooral in het
t:emperatuur gebied rond 1350 °C (dit is ongeveer de
bodemtemperatuur van de smeltwan). De vormingssnelheid van
2:uurstofbellen per mol procent van de oxyden die onderzocht zijn,
i.n afnemende volgorde is: Li 20, K20, Na20, BaO/SrO, CaO en MgO.
De diffusiesnelheden van de kationen van deze oxyden in het AZS
Em, daaruit volgend, het belvormingspotentieel inclusief de
invloed van de glassamenstelling en de temperatuur, zijn gebruikt
i.n een eenvoudig model van een industrHHe smeltwan met
realistische glassmelt/smeltgegoten AZS contacttemperaturen,
2:odat de belvormingspotentieel ten gevolge van de interactie van
smeltgegoten AZS en glassmelt berekend kan worden.
De verschillen in glasfoutpotentieel van de interactie tussen
smeltgegoten AZS en verschillende glassmelten berekend met deze
methode, komen overeen met de verschillen in de hoeveelheid
bellen en knobbels gevonden in verschillende industrieel
geproduceerde TV glazen.
De laboratorium experimenten tonen een stijging in aantal, in
zuurstofgehalte en in diameter van de gevormde bellen, bij
stijgende temperaturen (van 1350 naar 1500 ·C) .
De chemische samenstelling van de kern van de knobbels, afkomstig
van de overgangslaag tussen glassmelt en smeltgegoten AZS blijft
onveranderd gedurende het oplossen in de bulk van de glassmelt
tot net voordat de knobbel helemaal opgelost is. Deze
onveranderde chemische samenstelling maakt het mogelijk om de
ontstaanstemperatuur van de knobbels te bepalen, doordat de
chemische samenstelling van de overgangslaag voornamelijk bepaald
wordt door de chemische samenstelling en de temperatuur van de
glassmelt tijdens de interactie.
Dankwoord

Het onderzoek is verricht in het laboratorium van de Basic

Technology Glass (BTG) van Philips.
De leiding van de BTG dank ik voor de gegeven mogelijkheid tot
promotie op dit onderwerp.
Veel dank ben ik verschuldigd aan Ruud Beerkens die als co-
promotor het schrijven van dit proefschrift op stimulerende wijze
intensief heeft begeleid.
Prof. de Waal en Prof. van Loo dank ik voor hun begeleiding.
Het tot stand komen van dit proefschrift is mede mogelijk gemaakt
door medewerkers van de BTG. Mijn hartelijke dank gaat met name
uit naar: Ad Verbeek en Ivo Lemmens voor hun assistentie bij de
experimenten en het analyseren van de bellen, Corrie Jaarsma,
Hans KrUsemann en Brigitte van Dijk voor de SEM/EDX analyses en
Martien Hendriks en Leen Jongeling voor hun ondersteuning op
lichtmicroscopisch en materiaalkundig gebied en het identificeren
van de knobbels.
Mijn vrouw ben ik erkentelijk voor haar geduld en steun tijdens
het schrijven van het proefschrift.
CUrriculum vitae

Frans van Dijk werd in Leende geboren op 27 mei 1956. Vanaf 1968
tot 1972 volgde hij de MAVO opleiding, gevolgd door de studie
chemie aan de H.T.S. te Eindhoven. Vanaf 1977 studeerde hij
Scheikundige Technologie aan de Technische Universiteit te
Eindhoven. In 1982 slaagde hij cum laude voor het ingenieurs-
examen bij de vakgroep voor fysische technologie bij Prof.
Rietema.
Vanaf 1982 is hij werkzaam bij Philips glas, met als zwaartepunt
smelttechnologie. In deze periode heeft hij diverse functies
uitgeoefend o.a. modelontwikkelaar, fabriekstechnoloog smelten,
projectleider glasfouten onderzoeken en smelttechnoloog Basic
Technology Glass.
STELLINGEN

behorende bij het proefschrift van F.A.G. van Dijk

1 De redox van het glas wordt in de literatuur verschillend

gedefinieerd bijvoorbeeld als de verhouding Fe3+ /Fe 2 + of de
zuurstofdruk van de glassmelt. Deze definities zijn echter
niet volledig. Een meer volledige definitie van de redox van
het glas waarin behalve de beide genoemde, de concentratie van
de polyvalente ionen is opgenomen, zou meer duidelijkheid
scheppen.

Chopinet M.H. et.al., Glastech. Ber. 56K (1983), p596.

Beerkens R., Glastech. Ber. 63K (1990), p223.

2 De bij kamertemperatuur bepaalde C.O.D. waarde ter

karakterisering van de redoxtoestand van het grondstoffen
gemeng is niet zonder meer bruikbaar voor de bepaling van de
redoxtoestand van de uit dit gemeng vervaardigde glassmelt,
vanwege mogelijke redoxreacties die aIleen bij hoge
temperatuur plaatsvinden en de uitwisseling van zuurstof met
de ovenatmosfeer.

3 De oude glassmelters vuistregel "hoe heter hoe beter" gaat

niet in aIle gevallen op vanwege de temperatuur
afhankelijkheid van de vorming van bellen ten gevolge van de
glassmelt/vuurvast interactie.
Nemec L. and Muhlbauer M., Glastech. Ber. 54 (1981), p99.
Dit proefschrift: Hoofdstuk 4.
4 De eenvoudigste manier om produktie uitval ten gevolge van
glasfouten te voorkomen is gevonden door de firma RAYWARE uit
de UK. "This piece of fine glass is totally handmade in the
old world tradition. Seeds and bubbles sometimes occur in this
process. These are not flaws, but illustrate the beauty of a
totally handmade article."

5 De hypothese van Auerbach, dat zuurstofbellen die ontstaan bij

de interactie tussen de glassmelt en het smeltgegoten AZS,
het gevolg zijn van selectief transport in het AZS van
moleculair zuurstof is onjuist.
Auerbach A., Symposium sur l'Elaboration du Verre, p295,
Madrid 1973.
6 De beschikbare middelen voor research en ontwikkeling kunnen
efficienter ingezet worden indien vooraf de economische
haalbaarheid van het gestelde doel onderzocht wordt.
7 Voor glasovens waarin een glas met een hoog kalium gehalte
gesmolten wordt kan beter een smeltgegoten AZS gebruikt worden
waarbij het bewust toegevoegde natrium vervangen is door
kalium. Het aantal glasfouten tengevolge van de interactie
tussen de glassmelt en het smeltgegoten AZS materiaal wordt
hierdoor geringer veer deze glazen.

Dit proefschrift

8 De steeds strengere milieu eisen vormen in de glasindustrie

een belangrijke impuls voor de introductie van nieuwe
technologieen die in de toekomst ook het smeltpreces kunnen
verbeteren.

Technologisch Meerjarenplan Glasproducerende Industrie

1990-2010.

9 Een kenmerk voor een industrie op zlJn retour is een toename

van het aantal managers en een afname van het aantal technici.