LIQUID-PHASE SEPARATION IN GLAZES

FOR SINGLE-FIRED POROUS WALL TILE

by J. Apartsiv", M.V. Nufiez!", A. Moreno'?' , M! J . Orts'"?
'" Esmalglass, s.a. Villarreal
""Inst it uto de Tecnologia Ceramica. Universit at Jaume 1. Castellon.
Asociacion de Investi gaci on de la s Industri as Ceramicas. (AICE). Castellon (Es pana ).
ABSTRACT
Th e appearance of colour differences in fast single-fired porous wall tile with transparent or
practically t ransparent glaz e is one of the defect s considerably impairing the quality of the finished
product.
Th e sepa rati on of immiscible liquid phases duri ng cooling of the glaze in the kiln was shown to
be t he physico-chemical trans for mation responsible for changes in colour. The colour of these glazes
was shown to be very sensitive t o small changes in the composition and limited alterations in th e
cooling cycle of the ware in the kiln. A method is proposed, based on the measurements of th e
chr omat ic coor dinates of the glaze obtained by subje cti ng it to different thermal treatment, for
det ermini ng the sens it ivity of th e glaze to changes in colour by alterat ions in the cooling cycle.
1. INTRODUCTION
The ap pearance of colour differences in fast single-fired porous wall til e with virtually transparent
glaze is one of the defects considerably impai ring th e quality of the finished product.
Th e colour of these glazes has bee n shown industri ally to be extre mely sen sitive t o sma ll changes
in the compositi on and in the cooling cycle. Th e lat te r seems to indi cate th at a seri es of physico-
che mica l t r ansformations (immiscible liquid-phase separ at ion, crystallization of new phases) t akes
place durin g cooli ng ofthe ware in th e kiln, which appear macroscopically as a change in colour of
th e glaze.
Th is kind of glaze is characterized by beari ng a high percentage of alkaline earth oxides and zinc
oxide as well as a consi derable amount of boron oxide. These mate ri als may therefore be expect ed
t o show a great tendency t o separat e into immiscible liquids or even to devitrify crystalline phases,
as may be observed from the corresponding immiscibil ty diagrams (1) (2) (Fi gur e 1). Th ere is in fact
a broad immi scibil ity domain for th ese systems where many of the compositions used as glazes for
single-fir ed tiles ar e t o be found.
245
coO MQO
CoO
5 i 02
Zn O
ZnO
5,0
2 CoO
2. AIMS
Figure 1. Immiscibility diagrams of some ternary systems.
In this study the causes of changes in colour of th e glaze are analyzed with a view to acquiri ng
enough information to formulate new composit ions whi ch avoid or minimize this defect.
A seri es of experi ment s was therefore programmed to achieve the following aims:
i) To determine the nature of th e physico-che mical transformations responsible for the change
in colour of the glaze on cooling in order to elimin ate or minimize this defect.
ii) To determine the temperature interval in which these changes in colour take place at th e
greatest rate in or der to choose a suitable cooling cycle and establish the influence of this firing
stage on this phenomenon.
3. MATERIALS AND EXPERIMENTAL PROCEDURE
3.1 Materials
In order to carry out this study, a group of frits used in obtaining transparen t glazes over porous
bodies by fast single fir ing were utilized. Their chemical composi t ions are det ail ed in Figure 2.
Alth ough differ ences in composit ion among t he differ ent fri ts are to be obse rved, they all bear a gr eat
amount of eaO and ZnO,as well as BaO in some instances.
The sealing, t r ansformati on, and softeni ng temper at ure values of th ese fri ts are detailed in
Figures 3, 4 and 5. All present the typi cal values for frits used in si ngle-fired porous wall t ile
manufacture.
246
(Ofo)
70 ,....-- - - - - - - - - - - - - - - - - - ----;
40
30
20
10
o
10.3
CaO MgO Na
2
0 K p BaO ZnO
Figure 2. Interval of composi tions studied.
3.2 Experimental
3.2.1 Sample preparation
Glaze suspensions wer e prepared by grinding each frit in a laboratory ball mill with 5%kaolin and
0.3%C.M.C. Operatingconditions were as follows:Suspen sion density 1.65 g/cm", and2%grind oversize
fraction at 40 urn. Th e glaze suspension wa s sprayed ont o a single-fired porous r edware body whi ch
had previously been fir ed aU1OOC, while the amount of glaze applied, 0.7 g/crn, was held as constant
as possible.
3.2.2 Thermal treatment.
3.2.2.1 Non -isothermal experiments.
A seri es of expe riments was ru n with ea ch of th e frits, keeping th e firing cycle constant and
modifyi ng the kind of cooling in order to compare t he changes in colouring th e glaz es undergo during
th e cooling stage of t he firi ng operation and to determine the temperature interval in which this
phenomenon takes place at t he greatest r at e. Th e th ermal cycles used are shown in Figure 6. As may
be observed in this figure , each test specimen was fir ed up to 11OOC and cooled in the kiln according
to a slow (A) or a fast (B) cycle unt il reachinga cer tain t emperature. The test specimen was the n taken
out of the kiln and left to cool in ambi ent condi ti ons. The arrows in the figure indi cate t he
te mperat ur e at which removal from th e kiln t ook place.
Th e thickness of th e glaze layer and its chroma t ic coordinates (Sect ion 3.2.3) were determined in
the cold t est pieces. Th e test pieces wit h a markedly different thickness in the ir glaze layer from th e
pr eset val ue were r ejected.
247
25
20
15
10
5 -
o -
860 870 880 890 900 910 920 930
Figure 3. Seali ng temperatures of th e fri t s used.
25
20
15
10
5
610 620 630 640 650 660 670
Figure 4. Transfonnation temperatures of the frits used.
248
30 ,----- - - - - - - - - - - - - - - - - - - - - - -----,
25
20
15
10
5
0 -
740 760 780 800 820 8 40
Figure 5. Softening temperatures of the fti ts used.
400
2'
1000
3'
4'
5'
6'
7 8
800
t:
8'
H OC)
A
600
zoo
10 30

50 70
t Cmin)
Figure 6. Non-isot hermal t hermal treatment used. (The arrows indicate temperature and time at which
test pieces were removed from the kil n).
249
3.2.2.2 Isoth ermal experiments
A series ofisothennal experi ments was conduct ed to detennine the influence of te mperat ure on
th e rate at which changes in colour take place in the glaze.
The thermal t reatment carried out is schematized in Figure 7. It involved subjecti ng the fired t est
pieces taken from the kiln at llOOC to differing th ermal treatment.
110 0' C 1--------,
T
/ I
/ I
t (mlnut os)
Temp erat ura del
t ratamiento
i soter mo
Figure 7. Isothermal treatment method to determine the rate of change in colour of the glazes.
3.2.3 Determination of the change in colour in the glazes.
The changes in colour in each glaze , which take place on cooling of the ware, were quantified as
the difference between the values of th e chromatic coordinates of the glaze appli ed on th e test piece
removed from the kiln at llOOC and th e values corresponding to those involved in th e remaining
thermal treatment. As pointed out above (Section 3.2.1), glaze was applied on bodies previously fired
at 11OOC so that any possible change in colour of t he body would not affect the resul ts.
3.2.3.1 Determination of the chromatic coordinates of the glazed piece.
These were detennined by means ofa t ri st imulus colori meter, model LFM-3, supplied by the finn
Dr. Lange. The CIE 1976 (CIELAB) mea suring system was used. This is based on colour descripti on
by using three rectangular coordinates (L', a' and b" ), The relati on between these parameter s and
colour are schematized in Figure 8. In order to reduce experimental error, five measurements were
run on each test specimen and the mean value was computed.
250
..L- (bla nco)
-b'
( amar i llo )
- a'
( verde)
-b -
(a zul) - L (neg ro)
- a'
{rcic}
Figure 8. CIELAB system of chromat ic coordinate s. (green; light green; light yellow; yellowred; dirty red;
blue black; blue)
3.2.3.2 Ca lculation of colour di fferenc es.
The standard ASTM D-2244-85 method 0 was used.
According to thi s method, colour difference (6C)' between a certain colour (C') described by its
chromatic coordinates (L' , a' and b") and the reference colour (C:) with coordinat es (L: , a: and
b: ) is given by the equation:
6C' = [(6L* )2 + (6a' )2+ (6b*)' ]' f2
where:
6 L* = L* - Lo*
~ a = a a o
6b* = b* - boo
A gl azed piece fir ed at 11OOC and r emoved from th e kiln at th is te mpera t ure was taken as
r eference colour (Co"). Th e values for 6L*, 6a* and 6b* there fore indicate th e difference between the
values of the chr omat ic components white, red and blue ofa glazed piece and th ose of the reference
piece.
4. RESULTS AND DISCUSSION
4.1 Ch a nges in colour undergone by the glazes according to the cooling cycle they were
s ubject ed t o.
In Figures 9 and 10 the bounding colour difference values found for a glaze fir ed at 11OOC and
251
r emoved from the kiln at thi s temperature and the findings obtai ned on letting it cool i n side t he kiln
to ambient temperature are shown for the glazes st u die d, according to each of the two above-
ment ion ed cycl es (Figur e 6).
60 r-- - - - - - --- - - --- - - - - - - - - - ----,
40
20
o
- 20
+ 6 L ' (blanco )
- 40 L- --"'-"'----'-'''''''''---- ---J
Figure 9. Colour difference and chr omatic coordi nates and between th e glaze fired
and removed from the kiln at 1100C (vi rtually transparent) and the glazes afte r cooli ng by a slo w cycle
(type A). Cooling time from 1100 to 750' C was 25 min .
30 .--- - - - - - - - - - - - - - - - - - - - ---,
20
10
-10
-20
+ 6 L ' (blanco)
t ransparente
+68. '(roj o)
- 68.' (verde)
+6b '(amar illo)
- 6b'(azul)
-30 L- --'
Figur e 10. Colour differe nce and chromatic coordinates and bet ween the glaze fired
and remove d from the kil n at 1l00' C(virtually transparent ) and th e glazes after cooli ng by a fas t cycle
(type B). Cooling t ime from 1100 to 750' C was 3.5 min .
252
The following may be concluded on analysis of these figures:
The magnitude of the change in colour th e glaze undergoes on cooling decreases consid erably
when the cooling rate of th e ware increases.
All th e studied glaze s showed changes in colour to a greater or lesser extent on cooling. However ,
some performed better and even managed not to show any appreciable changes in colour when
cooled at high rates.
In each case th echange in colour was due to an incr ease in the blue (-t.b*)and white (t. L*) chromatic
component.
The change in colour one of th e st udied frits underwent when this glaze was cooled according to
cycle A, is plotted in Figure 11.
60 .-- - - - - - - - - - - - - - ---,
TlPO DE ENFRI AM IEN TO A
40
20
o
SJ
/
BL ANCO
il X 0

VERDE
-20
il l
\
- 6 -
+ 6 b '
AZUL
-E- 6 C '
*
- 40
1 10 0 10 5 0 10 0 0 950 90 0 8 50 800 750
TEMPERATURA (' G)
Figure 11. Vari ation of the differences between the values of the chromatic components of the glaze and
those corresponding to the reference piece with cooli ng time.
In this glaze the change in colour is obs erved t otake place in a temperature interval bet ween 1000
and 900 "C. Anal ogous findings wer e obtai ned for th e other glazes studi ed. although the temperature
interval in which the change in colour takes place at th e greatest rate and th e magnitude of the
t ransfor mation are different.
The above seems t o indicate that t he change in colour of th e glaze on cooling is due to immi scible
liqui d-phase separ ation and/or devitrificati on of crystalline phases. In fact, the maximum rate of
253
change in colour takes place at a temperature interval in which the glaze is fluid enough to allowthe
above-mentioned transformations to take place. On the contrary, at lowtemperatures viscosityofthe
glaze is so great th at such transformations do not take place at an appreciable rate. At high
temperatures, close to peak temperature, the thermodynamic tendency to nucleate is so slight that
although viscosity of the glaze is also low, no nuclei can form. (3) (4)
4.2 Kinetics of changes in colour of the glaze.
A series of isoth ermal experiments was conducted with one of the frits which showed a great
tendency to change in colour on cooling, according to the procedure described in Section 3.2.2.2.
In Figure 12 th e differences are plotted between the values of the chromatic coordinates of the
glazed pieces treated at different temperatures and dwell times and those ofthereference piece (taken
from th e kiln at llOOC), in order to determine the temperature interval in which changes in colour
take place at the greatest rate.
40
r-s min
30
0
1'5 min
+
t'10 min
*
t'15 min
20

L', L
10

,

-10
-20
-30
1100 1000 900 800
- 40 L- -'-- -'---- ---'- ----...l
700
Figure 12. Evolution of the chromatic components of the glaze with isothennal temperature treatment,
for different dwell ti mes.
254
In Figure 13, the evolution of the chromatic coordinates blue and whi te are plot ted wit h dwell
t imes for the te mperatures at which changes in colour develop at the greatest rate , in order to
determine the effect of te mper ature on t he r ate of change in colour of th e glaze.
40 r--------------------,
AZUL
o
- 10
- 30

T'900'C
BLANCO
30
+
T'950' C
"*
T'100 0 'C
ti L
20
10

f', b -20
20
15
10 5
- 40 '-- ---' ---l ---l --I
o
TIEMPO (MIN.)
Figure 13. Vari ati on of the chromatic coordinates of the glaze with dwell times at different temperatures.
On analyzing these plots , the following conclus ions may be drawn:
(i) The t emperature interval in which colour changes take place in the glaze at an appreciabl e
rate lies between 800 and 1050C (Figur e 12).
(i i) Th e te mperat ur e at whi ch th e colour difference of the piece is t he gr eatest (for both blue and
white), incr eases as dwell ti me decreases (Figure 12).
(iii) The white chromatic component ~ V increases as dwell ti me of th e glaze increases for all
the t emperatures tested (Figure 13). On th e cont rary, the blue chromat ic component r eaches
its maximum value after a certai n ti me and t hen drops . In fact for this glaze, at temperatures
close to 900C, maximum blueness is reached after a te n-minute dwell t ime. Blueness
decrea ses progressively and whiteness increases on incrementing dwell time at this
te mperat ure (Figure 13).
(iv) On comparing the evolutio n of t he chromatic compone nts with dwell ti me at a certai n
t emp erature (Figure 13), t he change in colors of the glazes is observed to followthe sequence:
255
Tran sparent - blue - bluis h white - white, in each case.
The fact that in both isothe rmal as well as non-isoth ermal treatment (Section 4.1), blu ene ss
appears as th e fir st observe d change in colour of the glaze, to the n tum white, indicates that the
physi co-chemi cal t r an sformation responsibl e for this phe nome non is immiscible liquid-phase
separation. In fact, blueness is due to mainly the shortest wavelength radiati on of th e visible
spectru m being absorbed by th e glaze. This absorption is characteris ti c for glazes bearing nucleated
phase separation in which th e size of the dr ops dispersed in the matrix is limi ted. On increasing the
durati on of th ermal treatment, dr op size grows and/or th ey coal esce, which entails absorption of
radi ation with a longer wavelength and t he glaze therefore appears whi t e and opaque (6).
4.3 Nature of the physico-chemical transformations responsible for changes in colour
of the glaze on cooling.
In the foregoing secti on, immi scible liquid-phase separation was suggested as being the likely
cause of changes in colour of the glaze on cooling. In or der t o confirm this assumption some
isothe rmally tre ated test pieces were chosen for observation by scanning electron microscopy (SEM).
Transpa rent glazes wer e shown to contain nucleated separation of immiscible liquid phases with
a very small drop size O.2Ilm) (Figure 14).
Furth ermore, blui sh transparent glazes (Figure 15 and 16) were also observe d to bear nucl eated
phase separ ation. However , drop size was great er (O. 2Ilm) so that on mainly absorbingshorter visible
wavelength radiation th ey turned bluish.
Figure 14. Microstructure of a transparent glaze.
256
Figure 15. Microstructure of a bluish glaze.
Figure 16. Microstruc ture of a bluish glaze.
Whi te (opaque) glazes were similarly observed to bear drops ofimmiscibl e liquid phase which had
coalesced (Figur es 17 and 18). The increase in size of the dispersed phase involves overall absorption
in the visibl e light spectrum thus giving the glaze a white appearance.
257
Figure 17. Microstructure of a bluish-white gl aze.
Figure 18. Microstructure ofa white glaze.
It may be concluded from th e foregoing that th e colour of th e glaze will dep end upon the number
of di spersed drops and th eir size. Th er efore, as nuclea ti on of small drops and their growth and/or
coales cence take place ata great rate and at partially overla pping te mperature intervals in this glaze,
holdin g the colour of th e glaze or periodically obtaining transparent slightly bluish glazes would be
ext r emely complic ated (Figure 19).
258
Intens i dad
del
colo r
I .... "
AZ UL
II
BL ANCO
TEMPERATURA ('C)
Figure 19. Nucleati on (I) and growth (Il) curves.
4.4 Isochromatic curves.
Th ese curves wer e experi me ntally obtained by mean s of isothermal techniques (Secti on 3.2.2.2)
and pl ot t he pair s of values , te mperature an d dwell t ime at t hat t emper ature, where the gl aze r eaches
the same blue ness (Figure 20) or whit eness index values (Figur e 21). Th e following informati on may
easily be acqui red by using th is kind of pl ot :
(i) The sensit ivity of a glaze t o change in colour on cooling. In fact, the shorter the t ime in whi ch
a certai n change of colour takes place the mor e diffi cult it becomes for the gla ze t o keep it s
ori gi nal colour on cooling.
(i i) Th e te mperat ure at which the change in colour ofthe glaze develops at th e greatest r ate. That
is, the temperature at which the glaze undergoes a certai n change in colour with the shor test
dwe ll time.
(iii) Est abl ish ing the t emperature and ti me intervals in which the r e is no change in glaze colour.
259

\
I
/
I

5 10 ZO
t (.I n.)
lot
. Zot
"'3D'l.
Z
--
/ ...----
/ /"
\. (
<,
900
TeC)
1000
1050 t-----::-:::-=;;r==::;::::::::---- ----J
Figure 20. Curves of the same blueness index values ~ b obtained from isothermal treatment.
1050
........ _k
...
/' ..- .


1000
... (
. ~
<,
'- .
--
<,
TeC)
900
"' st

. 1ot
dl
. Ist
"Zot
Z
5 10
ZO
t (.-i n. )
Figure 21. Curves of the same whiteness index values (l1L*) obtai ned from isothermal treatment.
On analyzing Figures 20 and 21 the following conclu sions may be dr awn:
(i) For this frit, th e greatest rate at which change in colour takes place lies at a temperature of
about 1000C. At this temperature a dwell ti me of 1 to 2 minutes leads to a considerable
increase in bl ueness and opacity (whiteness) of the fri t . At lower or highe r temperatures t han
the above the time req ui red to produce appreciable change in colour is longer. For temperatures
above 1050C or below 850C the rate at whi ch chan ges in colour develop is very low.
(ii) The temperature and ti me intervals in which this glaze shows a certain blueness and
whiteness are extremely narro w, thus rendering it ext remely complicated to hold the colour.
(i ii ) On superi mposi ng Figures 20 and 21, the areas in which the glaze turns blue virtually coincide
wit h those in which t he glaze shows opalescence. Th ese findings confirm the extreme difficulty
of obtaini ng t rans pa rent blue glaze s whi ch are not opalescent wit h this frit.
260
1000
900
S
H C I
800
o tlb"
",l Ot
6b ..20':.
db- .30l.
700
10 100 1000
Figure 22. Curves of the same blueness index values (t>b") for a glaze with little
tendency to li quid-phase separation.
1000
v
900
TC C I
800
700
10
6.L " ..S 't.
o 6L .10l.
6L . 15"'-
"" 6l 11; .. ZO':.
100
t ( min )
1000
Figure 23. Curves of the same whiteness index values (6 L*) obtained from isother-mal treatment.
It has bee n possible to obtai n frits for single-fired porous ti le with little te ndency to change in colour
on cooling, on the basi s ofthe influence exerted by the fri t composi ti on on liqui d-phase separati on as
well as by using th e te chniques described above.
Th e isochromat ic diagr ams correspondi ng to one of these Frit s are plot ted in Figures 22 and 23.
It has likewi se been observe d that slight var iat ions in frit composition considerably infl uence th e
261
tendency for changes in colour to develop on cooling of the glaze. This all entails th e need to dispose
of a very stri ct monitoring system both for th e raw material s as well as th e fri t elabora tion process
to obtain good transparent glazes.
5. CONCLUSIONS
The following conclusions may be drawn from the findings obtai ned in this study:
(i) Immi scible liquid-phase separation was shown to be the physico-chemical trans for mation
responsi ble for changes in colour of the glaze on cooling in th e kiln, for th is kind of glaze .
However, the magnitude of this phenomenon largely depends on frit composition and the
cooling rate used on firing.
(ii) On studyi ng the evolution of the chromatic coordinates of th e glazes subjected to different
ther mal tr eatment, change in colour was in each case shown to follow the sequence:
t ran sparent -blue - bluish whi te - white. The study of the microstruct ures of these glazes
showed these changes in colour to be associated with growth and/or coal escence ofi mmiscible
phases separa ted by nucl eation.
(iii) The isochroma tic diagrams described above, which pr ovide data concerning the sensitivity
ofthe glaze to changes in colour and the temperature at which the tra nsformation takes place
at the greatest rate, allow fri t compositions and the most suitable cooling cycles to be chosen
to avoid these chan ges in colour of the glaze . It is possible to obtain good trans parent glazes
for single-fired porous wall tiles by means of the se diagrams, with the usual cooling cycles.
6. REFERENCES
(1) Rinc6n, J . "Separaci6n de fases en vidrios", Ed. S.E.C.V. Madri d (1982)
(2) Vogel, W. "Chemist ry of Glass". Ed. Am. Cer. Soc. U.S.A. (1985)
(3) "ASTM Standards on Color and Appearance Measurement" Ed. ASTM Philadelphi a
(1987)
(4) G. GEIRNAERT, Verres et Refracta ires, 21-30, 25(1) (1971)
(5) E. PLUMAT, Silicates Industriels; 5-13,1 (1967)
(6) L. PROD'HOMME, Verres Refra ct ., 604-613,22(6) (1968)
262