CASTELL6 :-J (SPAL,\) is QUALI

2002
AVOIDING WATER MARK FORMATION BY
ADJUSTING ENGOBE LAYER CHARACTERISTICS
Fabio G, Melchiades'", Luciano 1. Silva"', Valdemir A. Silva''' ',
[oao Celso Romachell i"", Douglas D. T. Vargas'?", Anse lmo O. Boschi''
"T.aboratorio de Revestimentos Ceramicos - LaRC - DEMa - UFSCar.
Rod . Wash ingt on Lui z, Km 235, Cai xa Post al 676, CEp: 13565-906, Sao Carlos, 51' - Brazi l
e-mail: daobccpower.ufscar.br
("lCeramica Porto Ferrei ra Ltda.
Av, 24 de Outubro, nol , CEp: 13660-970, Port o Ferreira, Sp, Brazil
e-mail: sac-info@ceramicaportoierreira.com.br
""'Centro Ccrarnico do Brasil
Rua Machad o Bitten court, 205 - sala 86, CEp 04044-000, Vila Cleme ntine, Sao Paulo, SP,
Brazi l
e-mail: dougl astrian a@ccb.org.br
ABSTRACT
The water mark. a rclatirclv COll llllOlI pltellOlllelloll ill ceramic tiles, consist of all alteration
of the aesthetic properties of tiu: glazed tile suriacc tohcn the body becollies wet. Groen the grmt
number of claim: l11ade bccnuec or titis type or problcn: blf ceramic tile consumer ill Brazil. tlte
objective of thi stlldy ha been to cont ribute to minimisins; the pro/Jlel11 by stlldyi llg the effect of
the ellgobe composition 011 the water mark ill singlfire porotiS tiles. The evaluation or the result
cOllsisted of ntensurins; cltl1l lges ill colour of the glazed tile surj acc blf a colorimeter, after preset
periods of water innncrsion. Besides calltributiII g to a better understandillg olth phenomenon, the
rrsult indicate tluti tilt' ellgobe COI11 POSitiOIl plays a key role ill the dcreloinncnt of the 1m ter murk
11II d that its occtlrrell ce call !>e aroided by appropriate cngobeformula tion.
p, (; 1- -B 3
iI QUALI u 2002
1. INTRODUCTION
CASTELL6 "J (SPAl"J)
The term water mark is used to describe "darkening" of the gla zed ceramic tile
surface, whe n the porous tile body absorbs water. When thi s phenomenon appears with
alreadv installed tiles, the altera tion of the tile colour adverselv affects the aestheti c
charac"teristics of the environment. Wat er can enter int o con tact with ceramic tiles in
different ways:
mortar used in the ii xing;
intentional wetting of the tiles before iixing;
wa ter present in the external environment, which can enter through the joint;
pipe leaks in internal building structures;
moisture from wa lls with inadequate waterproofing;
tile exposure to moist environments before fixing.
Wat er marks are currently the main source of claims and lawsuits involving tiles
certi fied in Brazi l. An initial ana lys is carried out by the Ceramic Centre of Brazil (CCB) -
organisa tion responsible ior ceramic tile certi fication in the country - indicat ed that of 496
certified product s made by diiferent manufacturers, approximately 40% of these
presented wat er marks. It is impor tant to point out that the grea t majority of Brazil ian
consumers seeks glossy, transparent products, where water marks are much more likel y
to appe ar.
The res ults of statistica l analys is of the data collected by CCBare presented in Figure
1. It can be observed that wa ter marks preferentially occur in products with hi gh apparent
poros ity, pa rt icul arl y in semi-porous and porous tiles.
60
50,8
t
50
to
33.3
~
-lO
'" 30
"
20
"
~
10
BID BUb
2,2 0,0
SUa e u
Water absorptionclasses
Figure 1. Effect of the absorption class all the ilppearmlce ~ tile imler //l ark.
In order to effectivel y eliminate the problem, or at least contro l it, it is necessa ry to
under st and its origin. The evidence suggest s that the origin of the water mark is relat ed
P. LI-Bo
CASTELLC) N lSPA Il'\ )
:1'. QUALI
... 2002
to the vis ua lisa tion, th rou gh the gla zed surface, of the altera tion of the visual appea rance
of the porous bod y on becoming wet. The refore water marks only occur when the
comb ined opacitv' " of the glaze I engobe has not been su fficient to impede visualising the
cha nge in "colour" of the body. Thus, cons idering that the glaze will conti nue to be
transpar ent and the body porous, the only feasible alternative for solving the problem is
by controlling the opacity of the engobe!". Engobe opacity can be considered to depend
fundamentally on three par amet er s: 1) thickness of the layer; 2) tr ans parency of the solid
phase and 3) presen ce of open pores. Points I and 2 are self-exp lanatory but the effect of
porosity deserves to be analysed in more detail .
Briefl y, with regar d to the issue at point the key fea ture is to understand the
d ifferen ce between wha t happens when the pores are full of air or wat er. It is well known
that pa rt of the incident light on an interface bet ween two tra nsparent media, depending
on the angle of incidence and difference bet ween the refractive indexes, can undergo "
process known as total int ernal reflection (TIR). The conditions in which TIR occurs are
fou nd from the following equa tion :
[II
Wh ere 0, is the cr itical angle from which TIR is given (angles of inc idence larger
than 0, wi ll undergo TIR) and n
l
and n, corre spond respecti vel y to the ref ractive indexes
of the media from which the light comes and through whi ch it passes. As consequence of
the TIR phenomenon , pa rt of the incid ent light does not penet rat e inside the pore.
Cons idering n, IM' = 1.5, n ", = 1.0 and '1" ."" = 1.3, it can be concl uded from eq ua tion [11that
the cr itica l angles of incid en ce (8) for pores filled with air and wat er are 41.8
0
and 50.3,
respecti vel y, i.e., the rays of light whose angles of incidence on the glaze l pore int erface
are la rge r than these va lues will undergo total internal reflection (TIR) and not penetrate
ins ide the pores. Thus the quantity of light that penetrates inside the pores of the engobe
is larger when these are filled with wa ter than with air. Therefore, as the engobe is
porous , when the pores are filled with wate r, they become more transparent than whe n
they are filled wit h air.
Ano ther way of approaching the prob lem is by evaluating the fraction of incid ent
light (I E) that would be re flected (IR) by an int er face bet ween two mat erials with different
refracti ve indexes (lI"and 11), according to the equa tion [2].
I _ I
E
( II" - 11, ) '
, -
(11" + 11, )'
[21
Using the refracti ve ind exes gi ven previo usl y, the conclusion is rea ched that the
int ensities of reflected light (I , ) for the air I glass and water I glass interfaces would be 0.04
and 0.0051 respecti vel y, as illu strated in Figure 2.
1' . C l - 4.'7
II QUALI ' L/? 2002
Dry Tile
CASTEl.1.6 N (SPAIN)
Figure 2 ltlus tration of (Jl'llcity by thrporous aftcr il' ctt ill,'\ .
This preliminary ana lysis indicates that from the point of view of the engobe, the re
are basicall y three alternati ves for address ing the problem of wat er marks: I) increasing
the thickness of the layer; 2) increasi ng the opaci ty of the solid phase and 3) eliminating
open pores. These alternatives could be use d separat ely or simultaneous ly. It is to be
noted that, in relati on to othe r aspects, it is more convenient and therefore more pr act ical
for man ufacturer s to use porous engobes.
Other alternatives, not related directly to the engobe, would be: 1) red ucing porosity
of the bod y and 2) increasing gla ze opacity.
Engobe layer thickness is controlled by adjusting application cond itions, and as it is
economically feas ible and does not affect othe r characteristics, this can be easil y done.
The solid (non-porous ) phase of the engobe usually consists of a gla ssy matri x
(amorpho us) and crystalline phases. The opacity of this set of ma terials depends on the
differences bet ween the refracti ve ind exes of the glassy matrix and the crys tals].']. As
shown in equa tion 2, the greater this difference, the greater will be the opacity of the set
of mat er ials. Crystals can be added to the engobe formula or form during firing. At
present, zirconium silicate is the most widely used opacifier owi ng to its high refractive
index (1.94) and high refractoriness, which is a fundamental requirement so that the
crystals do not dissolve in the glassy mat rix duri ng firing. Table I presen ts the refractive
indexes of some materi als.
Materials n
Glass 1.50
Mullite 1.65
Alumina 1.76
Zirconiumsilicate 1.94
Air 1.00
Water 1.30
Table I. Refrnctiiv indexes ( 11 ) of some materials.
P. (;1 .
CASTELLO\! (SI'A ['4 )
i! QUALI 2002
Open pores can be eliminated by increasing engobe meltability. The fluxes used in
the engobe formulas can be divided into p ri mary and seconda ry fluxes!", In the p rimary
fluxes, white frits are the most noteworthy, which soften at temperatures lower than 800
0
e
and star t the d isso lving process of the most ref ract ory components. Raw materials wit h a
low melti ng point or their eutectic combi na tions ar e used as secondary flu xes. Feldspa rs,
talc and the raw ma terials that contribu te alkaline and / or alkalin e-earth elemen ts are
found in this group.
2. OBJECTIVE
The obj ective of the present work has been to es tablish the main aspects to be
considered in defini ng the engobe composi tion tha t would enable eliminating water
marks in porous ce ra mic tiles with transpar ent glazes manu factu red by single firing.
3. EXPERIMENTAL PROCEDURE
The ex pe rimental procedure adopted was d ivi d ed int o two stages. Th e objecti ve of
the fir st stage was to deve lop a simple and effective tool for quantitati ve ly evaluating the
formation and di sappearance of water ma rks. Using the met hodology devel oped in the
firs t stage and the analys is set out in th e introducti on, the object ive of the second stage
was to st udy the effects of the following variables: 1) layer th ickness; 2) transparency of
the solid ph ase and 3) presence of open pores.
An interes ting feature of the methodology used is its contribution to a be tt er
underst anding of the role of the engobe which, though found in almost a ll porous tiles,
has till today p racti cal ly never been systema tically st udi ed in d epth.
3.1. QUA;\;TI FICATtON OF THE WATER NIARK
Although the few works published!' ! on water marks su ggest the use of colorimetric
techniques to measure the intensity of the variation in colou r p resented by the tiles on
absorbing water, none go into further details concerning this possibi lity. In view of th is, it
was decided to devel op a met hodology based on th is s uggestion. Taki ng into accou nt the
above rema rks, it was decided to characterise the phenomeno n by eva luating th ree
fundament al aspects: 1) wa ter mark formati on rate, 2) maxi mum intensi ty after a certain
period of wate r immersion and 3) water mark disappearance rate.
Th e adop ted experimental proced u re is schemati cally presented in Figu re 3.
Ini ti all y a reading was made using the three col our co-ord ina tes of three dry tile regions,
with a portable colorimeter (Minolta model Th e th ree meas urement point s
were defined by 111CanS of a template to ensure that after water immersion, the readings
would be made a t the sa me points . Having made th e th ree readi ngs th e mean values of
the co-ord ina tes !":', L*, a" and b", su pplied by the co lorimet er in each reading, were
calcu lated. After wate r immersion for p reset ti mes, the glazed su rfaces were d ried wit h
a d ry clo th and the colorimetric readings were pe rformed at the same poin ts as before on
the d ry tiles . Th e measured colour variatio ns fou nd in the tile s befor e and after water
immersi on were eva lua ted by means o f the para met er AE*, which allowed quan ti fyi ng
the intensi ty of the arising stains.
j '. GI - D Y
iI QUALI ~
CASTELL6 l': (SPAIN)
To evaluate the di sappearance of the water mark a similar procedure was adopted,
the onl y difference being that the periods of time between one colorimetric measurement
and the next one, now corresponded to drying time in air.
In the study conducted, measurements were made on the tiles after every 5 minutes'
immers ion, seeking to determine the kinetics of mark formati on. After an hour' s wat er
immer sion, the tile was set to dry in a ventilated environme nt, keeping the faces exposed
to air. Colorimetric measurements were periodically carr ied ou t on the tile, always at the
three points initi all y established with a template, with the intention of evaluating mark
disappearan ce.
Water Tile+
Inmersion Tile Template Template
~ = > n O=>D=>L.. a. h.
~ 000
Water mark
formation
.:lE*
- -- - - - - - ---,
: Disappearance
.1..E* : of the water mark
Tile+
nl'oemPla,e=> l' DemP late=> L:a.. b..
00 0
Tile+
r l'o emPlate=> l'oe mPlate=> L. a. b
000 000
Time
Initial characterisation
Tile
Figure 3. Schematic illustration of the procedure used to em!uatl' the wafer mark.
3.2. STUDY OF TH E VARIABLES THAT AFFECT WATER MARK FORc\1ATION
First the effect of va ryi ng engobe layer thickness was evaluated . In this part of the
study a standa rd engobe was used for monoporosa products that presented a hi gh
tendency to form wate r marks. The engobe was milled in a fast laborat ory mill until
reaching a reject be tween 0.5 and 1.0%, in an ABNT ;;325 sieve (aperture of 44 urn ).
Subsequently, engobe layer s we re applied of 40, 50, 60, 70 and 80 g on porou s industrial
tiles of nominal dimensions 30 x 40 em. Such applications corres pond to specific layers of
330, 420, 500, 580 and 670 g we t engobe / m- tile . The tiles were glazed and fired in an
industrial kiln at 1120C in a 45 min cycle. Tile water absorption after the firing was
approximately 15.0%.
To evaluate the effects of the engobe composition on water marks, several formulas
were developed, using different raw materi als in va rying proportions. Initiall y two
engobes were us ed. The first, called standard engobe (P) is mentioned in the literature as
a typical example of a porous wall tile engobe. The second was devel oped from the
standard engobe (P), reducing the refractory raw mat erials content and rai sing flux
content. As this composition is more fluxing it is referenced F. Table II sets out the
compos itions of P and F
1'. CI - -l-lO
CASTELL6:--: (SPAI")
Raw materials ( 0/0) P F
Aruemina clav I 18.0 12. 0
Kaolin 6.0 3.0
Bentonite 004 004
Quartz 18.0 10.0
Sodi um feldspar 19.6 36.6
White frit 20.0 20. 0
Zirconium silicate #500 IH.O 18.0
Tctrusodi um 004 004
pvr ophosphate
CMC 0.2 0.2
Based on the more fl uxing composition (F), others were developed that allowed us
to study se parately the influence of each raw ma terial on the characteri sti cs of the cugobe.
The effects of the following cha nges in the composition of this engobe were evaluated:
Na ture of the clay;
Zirconium silicate pa rticl e size and content;
Content of the raw material s that con tri but e alkaline-earth oxides (talc, d iopsid e
and wollastonite);
Nature and content of the feldspars (orthoclase, albite and s pod u rn ene):
Frit content.
In this stage of the study, processing cond itions we re kept steady. Engobe reject was
held be tween 0.5 and 1.0% on an ABNT ; 325 sieve (44 urn), while engobe an d glaze layer
application was controlled in terms of the a pplied weight, at 625 g engobe / rn- and 735 g
gl aze / m' . Firing was conducted un der the sa me cond itions descr ibed previous ly,
The glaze used was al way s the sa me one, unless ex plicitly changed, and
corresponds to an industrial singl e-fire glaze essentia lly consisting of frits and sus pe nding
agents . In the fina l stage of the work the effects of the foll owing modificat ion in its
chemical composition were evaluated:
Substit ution of the tran sparent glaze by an opa que glaze;
Addition of d ifferent white frit contents to the s tandard glaze;
Addi tion of d ifferent zirconium si licat e contents to the st andard glaze.
All the tests carried out were ana lysed quantitativel y rega rdi ng wat er mark
for mati on and disappearance by colorimet ry, in accorda nce wi th the procedures set out in
point 2.1.
4. RESULTS AND DISCUSSION
4.1. Q UA0: TITATIVE EVALUATIO N O F THE WATER Ivl i\ RK
Figure 4 plot s the evolut ion of the water mark with wa ter imme rsion time and the
disappearance of the water mar k during tile d rying. The curve shows tha t the int e nsit y of
P. CI -
CASTELL6:--J (SPAIN)
the stain increases qui ckl y at the beginning and later tends to stabilise. Tile stai ning
develops practica lly during the first five minutes of immer sion and then exhibits a low
rat e of growth. During drying, the mark is observed to di sappear gra dually, presenting a
slightly higher drying rate at drying start. The time required for the elimina tio n of the
stain was cons iderably lon ger than the time needed for it to form. Analysis of these results
indicat es that the developed methodology enables quantitati ve eva luation of the
evolu tion of stain intensity across time and is shown to be appropriate for the purposes of
the present work. The gene ra lly used visual evaluations only permit compara tive
assessments.
...0
2.5
0
..0-
2.0
15
<
1.0
0.5
0.0 J
0 10 zo JO

Staining lime (min)
50
' .0
3,5
r.o
2.5
2.0
<
1.5
1.0
0.5
0.0
o
Dryin g time (min t
Figure 4. Variation of stahl intellsity (.1 *) with time.
2000 1500 1000 500 o
Marl<
-!----,O"-", intensity
4.0
3.5
3.0
2.5
2.0
Cl
-e
1.5
1.0
0.5
0.0
-0.5
Tune (mi n)
Figure 5. Stail1il1g Clin'c of all indust rial tile with thejonnation of a l1igll intensity tcater mark.
Combining the curves of water mark formation and disappearan ce generates the
staining curve (Figure 5). The maximum point of the curve represents the maximum
intensity of the mark after water immersion time. The curve slopes in the staining and
drying stages respecti vel y indicate the staining and di sappearance rat es.
Ana lys is of the results found in the later phases of the work confirmed the existence
of a direct relation between mark intensit v at the end of the wa ter immersion time and its
intensity after 3 hour drying (Figure 6). This res ult indicat es that for a body mad e from
the sa me composition, under the same processing condi tions, water mark drying rat e
does no t alter very much, inde pendently of the engobe and glaze used. Therefore, the
results set out bel ow only take into consideration the intensity of the mark formed after
30 min immersion.
P. CI - +12
CASTELI.6 r\ 2002
:'.0
"'.5
0
HI
ec
.'5
s 0
a
.f
0 0
0
x:
_'.0
0<9 0 0
-r,
ifi 0
0

0 0
0
. ::.0
'<
15
cEo00
c 00
1.11
05
, ,
.:1E* aft.. -r .'\() min. inmersion
4.2. STUDY OF THE VARIABLES THAT AFFECT WATER NI ARK FOR\IATI ON
4.2.1 Ellgobe Illyer thickness
Figure 7 s imultaneously plots the staining curves of the ti les made with d ifferent
engobe layer thi cknesses, as well as the rel ation be tween wa ter mark intensity and
a pplied layer wei ght. As mentioned previousl y, the st aining and mark di sappearance
ra tes did not undergo si gnifi cant alterations among the co mposi tions . Howe ver,
maxi mum mark intensity was noti ceabl y affected bv the variation of engobe layer
th ickness.
Figu re 7 sho ws that the tiles mad e with thi cker engobe layers presented less intense
marks than the othe rs. Thi s result indicat es that the thi ckness of the engobe layer can be
an important variable in avoi ding wat er mark formati on in porous tiles.
o
o
" ,--- - ------------,
j
H .
: 0
42 J
,
:::
3
E 3 8

g 3 6



Engobe layer (g/m
1
)
n ,)11 _lil l
""""- .<"l I Jl. m"
-O-- ",lI !1- m"
-4- " " JI.:m
Tin'll: uuim
ac
o
o
f igll rt' 7. qrccf llf c'I S0[JC layer (II I till' umtcr mark i l l
CASTELL6 !': (SPAl!':)
4.2.2 Ellgobe composi t ion
Figure 8 pl ots the results relating to the arising marks using engobe P and F, whose
compositions ar e detailed in Tabl e II. The increase in the meltability prov ided by replacing
the refractory components (clay, kaolin and qua rtz) by sodi um feldspar is observed to
affect water mark formati on, reduci ng mar k intensity. However, the product ma de with
engobe F still presen ts a visually noti ceabl e stain, verified by the hi gh t>E* (2.53) after
water immersion.
Table III gives the color imetric ana lysis dat a on the dry tiles, showing that the
increase in engobe meltability, besides reducing the water mark also produces a change in
colour of the product, ev idenced by the variation observed in chroma tic co-ordi na tes L*
and b*.
Composition L* a* b*
Enaobc P 90.4 0.5 1 5. 18
Enaobc F 89.7 0.52 4.00
Table 111. Colorimetric alzalysis of the products
made with composition s P and F.
35
3.0
~
*
2.0
OJ
< I ~
1.0
O ~
0.0
P F
Engobes
Figure 8. 1-Vater lJIark i ll tcll si t.1f of ctlgobcs P and F.
The effect produce d by replacing the clay used in the engobe composition was also
evaluated. These test s were conducted using composition F (Table II), whi ch had 12.0%
clay. Tests were carr ied out with four different clays in composi tion F, subseque ntly
assessing the intensit y of the water mar k forme d in the resulting products (Figure 9) and
the arising altera tions in colour (Table IV). It can be noted that the nature of the clay used
affects colour and engobe staini ng. Clays can affect the porosity of the engobe layer, both
by the packing that they contribute to the composition during the application and because
of their meltability in firing. The findi ngs confirm the need to use appropriate clays for
engobe development.
In the following test s, the indi vidual effect of the introduction or wi thdrawal of a
given raw materi al was evaluated, always holding the proportion of other ingredients.
First we present the resu lts found on adding and decreasing the zirconium silicate content
in composition F. Zirconi um silicate was used in two different grain sizes - ::325 an d #500
- with a view to evaluating the effect of particle size . Holding the proporti on of the other
raw ma terials, the ::500 zirconium silicate content was varied between 15.0 and 31.0%. For
the #325 zirconium silicate the variation was smaller (between 15.0 and 25.0%) in view of
the results found (Figure 10 and Table V).
P. G1 ~
CASTELLO'J (SPAl 'J )
Comoosition I L* I a* b*
F - Arucntina clav i 89.7 10.52 4.00
F - Paraiba clav 190.9 0.54 -l.t i
F- Suzano cl av 191.2 0.32 4.90
F - Enuland clav I 86.2 I 0.24 I 3.46
.,
, .U
35

. 1'<
"<
,
10
W

U.U
!!l QUAlI
England
Tl lNc IV - E.tfi.'Ct flzt' day,;, 0/1 t'lIgt1be
cotour.
Fig Hre 9- Effect of tire clay.;, 011 imta markill!tI/:, ity.
Zirconium silicate is a reiractory ingredient incl uded in engobe compositions to
provi de the necessary opacity!", Excessive zirconium silicate contents make the engobe
undul y ref ractory, thus raising the tendency to form water marks. The findi ngs indi cat e
the existence of a limit for the use of zirconium silicate, which depends on its initia l
particle size. In the case of =325 zirconium silicate, adding less than 20.0% to the
composition cont ributed to a reduc tion in wa ter mar k intensit y. Above thi s content, the
engobe becomes extremely refractory. The same result applies for ::500 zirconium silica te,
although in this case the limit for its use is much hi gher, owing to the finer particles, Of
all the zirconium silica te contents tested in the engobe composi tion, the results obtained
with "500 zirconium silicate were be tter than with ::325. This is due to the greilter
opacifying strength of the smaller particles and lower refractoriness that they contribute
to the composition. With regard to the color imetric analysis of the compositions, it was
observed tha t =500 zirconium silicate was ab le to provide smaller mark intensity without
ad versel y affecting engobe whiteness. In the case of =325 zirconium si licate, whiteness
decreased as a result of the iron impurities contained in the commercial raw materi al used
in the studv,
The foregoing results confirm that opacity is not eno ugh to avoid the formati on of
water marks. It is necess ary to have a ba lance between engobe opaci ty and meltability,
becau se the pores present in its mic rostructure can radicall y reduce opacity when the tiles
enter into cont act wi th water.
To st udy the int1uence of the t1uxing raw materials present in the composition, the
contents were varied of the frit , feldspa rs and raw materials sources of CaO and MgO
(talc, di opside and wollastonite).
Figure 11 and Table VI set out the results found with the va riation of frit content. All
the results refer to the introduction of a zirconium white fri t with a high coefficient of
expans ion and high softening point. It can be observed that addi ng the frit st rik ingly
raises composition meltability. Good res ults we re achieved with additions of up to 25.0%
whi te frit in the engobe. Above this val ue, the glazed tile surface bega n to deteri ora te and
the excess glassy phase began to reduce engobe opacity. It wa s also observed that the
presence of hi gh frit contents produces decisive alterations in engobe colour, wi th
conside rable loss of whi teness.
1' . CI 145
II QUAU c/./1 2002 CASTELL6" (SPAIN)
D
D
18 20 22 24 26 28 30 32
Zirconium silicate content (%)
o
o
o
o
55
5 0
e.s
E
=>
E
40
'x
'" E
W
3.5
-c
3.0
2.5
"
te
Composition L* a* b*
Enuobe F 89.7 I 0.52 4.00
F I \ 5% Zirconium S. #325 89.2 I 0.47 4.42
F I \ 8% Zirconium S. #325 89.4 I 0.37 4.52
F I 2 \% Zirconium S. #325 87.0 I 1.52 4.52
F I 23% Zirconium S. #325 87.1 I 1.37 6.34
F I 25% Zirconium S. #325 86.5 I 1.44 6.3\
F I \5% Zirconium S. #500 89.8 I 0.29 4.29
F I \ 8% Zirconium S. #500 89.7 0.52 4.00
F I 2\ % Zirconium S. #500 90.6 0.28 4.29
F I 23% Zirconium S. #500 91.0 I 0.27 I 4.49
F / 25% Zirconium S. #500 90.7 0.26 4.23
F / 28% Zirconium S. #500 91.2 0.03 4.68
F / 3\ % Zirconium S. #500 89.8 0.25 3.99
Tabh' v: Effcct of zirconi um silicate particle size fi nd
cOll le ll t 011 eJlgoh' colour..
FigureItt. Eifect oizirconiinn silicate particle:::. i: t' and
content 0 11 tlu: wala mark.
Comoosition L* a* b*
Enuobe F 89.7 0.52 4.00
F + 3.0% White frit 88.9 0.35 3.87
F + 5.0% White frit 88.8 0.21 3.72
F + 7.0% White frit 88.3 0.51 3.71
F + \0.0% White frit 88.5 0.13 3.01
F + 13.0% White frit 88. 1 0.13 2.59
2.'
2.'
D
2.2
D
E
X
10 2.0
E
W
t a
<
16
D
18 W u M
White frit content (%)
Table VI. Efft'ct o.{ wJz itejrif content Oil eJlsobe cololfr. Figure 11. Effect oficizitefrit content Oil tilt' willa mark.
Wit h regard to the results obtai ned with the increase in alkaline feldspar content in
composition F, which already contai ned 36.6% sodium feldspa r (Table II), the resu lts
obtained in respect of the objectives of this work were no t good. The analysed feldspars
(orthoclase, albite and spodumene) did not cont ribute to reduci ng stai ni ng and produced
little altera tion of engobe F colou r. Spodume ne differed slightly from the ot hers by
causing a certain drop in engobe whit eness - probably relat ed to the prese nce of
chromo phore elements - and a slight reduction in maximum mark intensi ty on addi ng
high contents to the standa rd composition (5.0 and 7.0%). Figure 12 an d Table VII detail
the results found for the studied feldspar s.
The raw materials sources of CaO and MgO were studied by virtue of the eutectic
composi tions that they can form when they ar e combi ned with alka line feldspars. The
study involved talc (magnesi um silicate), wollaston ite (calcium silicate) and diopside
(calcium silicate and magnesium), which had low chromopho re element contents and
contributed CaO and ! or MgO. The best resu lts were obtai ned by introduci ng talc into
composition F. Talc helps for m glassy ph ase, reducing engobe layer porosity. In thi s sense,
it can be conside red an importan t raw material for the development of imperviou s
engobes, although it brings with it some negative effects with regard to the whit eness of
the composition. Wollastonite and di opside did not produce the expec ted result, when
they were incorporat ed into composition F. Figure 13 and Table V11I set out the results.
P.GI- +J6
CASTEU6'J (SI'AI")
Composition L* a* b*
Enaobe F 89.7 0.52 4.00
F 3.0% Orthoclase 90.1 0.31 4. 17
F - 5.0% Orthoclase 90.3 0.42 4. 18
F - 7.0% Orthoclase 90.2 I 0.30 4.28
F - 3.0% Albit e 90.0 0.4 1 4.0 1
F 5.0% Albite 90.3 I 0.26 3 99
F - 7.0% Albit e 90.3 I 0.35 4.05
F -+- 3.0% Spodurnc nc 89.0
I
0.35 4.03
F + 5.0% Spodurnene 89.3 I 0.43 4. 10
F + 7.0% Spodumene 89.0 0.51 4.33
iI QUALI
a
-o--Spodume ne
-co-co nnccraee
"
--l:r- Albite
32
E
0
E a o
0
x
0
"E
28
OJ
<
26
"
0
2.2
o
2002
Tabte Vll . E. (fcct of nlknline feldspar content 0 11 l' Jl s obe
coknsr.
Feldspar addition to Engobe F (%)
Figure 11. Ejl ect t"Ol1tCll t on the u-o tcr
mark.
Composition L* a* b*
Enao be F 89.7 0.52 4.00
F + 3.0% Talc 88.4 0.40 3.99
F + 5.0% Talc 86. 5 0.23 3.28
F -+- Talc 87.1 0.48 3.79
F + 10.0% Talc 88.4 0.06 4.07
F 13.0% Talc 89.0 0.00 4.57
F Wollastonite 89.3 0.37 3.89
F 5.0% Wo llastonit e 89.1 0.28
I
3.69
F - 7.0% Wollastoni te 88.9 0.09 3.6 1
F - 3.0% Dionsidc 89.2 0.29 4.12
F 5.0% Diopside 89.2 0.30 4. 18
F - 7.0% Diopside 89.0 0.39 3.78
"
' .0
2.'
E
"
0
E 2'
x
re
2.2
E
W
2 0
-:
t.e
t.e
" -a
-o-Talc
-o-Diopside
---.:::r- Wollastonite
o
o
o
10 '2 14
Addition to Engobe F (%)
Ta/J le lITlI. E.fl eet talc. ii'ollastollite and diopside content
011 t'llgobecolour.
Figure 13. Effect lJtalc. trollastcnite and diop::.i de
Clmterlt 011 fhl! zmter mark:
Th e results indicate that additions of white Irit, talc and zirconiu m silica te contribu te
to red uc ing the water mark in porous ti le engobes. To conclude the studies on the engobe
composition, simultaneous addi tions were made of fri t, ta lc and zirconium silicate to
engobe F, al ways maintai ning the same proportions in the addition of these three raw
materials. As the results detai led in Figure 14 and Table IX show, the additi on of these
th ree raw materi al s reallv contributes to the reduction of the water mark wit hou t
significantly affecting cngobe colour, with con ten ts u p to 15.0% (5.0% frit, 5.0% ta lc and
5.0% ,\'500 zirconium silicate).
After performing numerous test s, some engobe compositions were selected that
presented th e best results, combi ning whiteness, rel ati vel y low or imperceptible
maximum water mark intensity and good gl azed sur face texture. Th ese engobes
p resented compa tible resul ts with the body and processing conditions used. However, the
succes s of these compositions for other conditions cannot be ensured . The main
contribution of the res ults is in the identification of the effects produced by ea ch raw
material, with regard to the reduc tion or increase of maximum water mark intensity.
Nevertheless, an optimum composi tion should be developed for each s pecific case,
considering the characteristics of the body; glaze and p roducti on process. The selected
engobe compos itions and the res ults of their charact erisation are set out respectivel y in
Tables X and XI.
1'. CI - 447
CASTELL6N (SPAI:',!)
-o-Mlx!ure 1:1:1
Frit + Talc + Zirconium sil icate
ComoositioD L* a* b*
Engobe F 89.7 0.52 4.00
F + 7.0% FTZ' 89.8 0.58 4.34
F + 12.0% FTZ' 88. 7 0.69 3.69
F + 15.0% FTZ* 88.7 0.57 3.60
F + 21.0% FTZ* 84. 7 0.56 3.07
*FTZ = [rit +talc + ff 500 zirconium siliciite (1:1:1)
2.6
2'
E
2.2
~
E
2.0
'x
m
E
W
' .6
<
' .6
"
' .2
0 5
o
to
o
o
is 20 25
Table IX. Effect offri f, talc and zirconium silicate
content 0 1/ cl1gobl' col oll r.
Raw materials (%) EI E2 E3
Argentina clay 10.5 11.2 11.4
Kaolin 2.6 2.8 2.8
Bentonite 0.4 0.4 0.4
Quartz 8.5 9.3 9.6
Sodium feldspar 31.5 34.0 34.8
\Vhite frit 22.0 18.8 23.8
Zi rconiu m sil icate #500 20.0 17.0 17.2
Talc 4.5 6.5
-
CMC 0.2 0.2 0.2
Tetra-Na pyrophosphate 0.4 0.4 0.4
Table X. Engobecompositions wi t11the best results.
4.2.3 Glaze compositi on
Adition to Engobe F (%)
Figure 14. Effect of jri f, talc and zirconium silicate
content 011 the water mark.
Characteristics EI E2 E3
L* 88.7 87. 1 88.8
a* 0.57 0.48 0.21
b* 3.60 3.79 3.72
hE* max 1.36 1.50 1.43
a O n ~ (x10-
7
"el ) 73.6 73.8 74.9
Table Xl. Characteristics of tile best compositions.
In cert ain cases where the composition and the thi ckness of the engobe layer are not
able to kee p water mar ks from ar ising, the re are some alternatives relating to the glaze
compos ition for minimising the problem. In thi s sense it is necessary to increase glaze
layer opacity so that the change in colour of the body is not very noti ceable when
observed from the glazed tile surface.
Figure 15 and Table XII set out the results found on repl acing a transparent glaze by
an opaque white glaze. The test s were car ried out on the same engobe composition
(Engobe F), and it serves to indi cate the effect of glaze opacity. The opaque glaze ena bled
considerably reduci ng water mark intensity, while the whiteness of the resulting produc t
increased. Thus, opaque glazes can be used successfully to minimise water marks, when
this is compatible with the design sought.
Finall y, wi th a view to evaluating the possibiliti es of reduci ng water mark intensi ty
in products that use tran sparent glazes, the possibilit y was tes ted of inc reasing their
opacity slightly by incorporating zirconium silicate and opa que white frit in the
transparent glaze composition. The results are presented in Table XIII and Figure 16,
indicati ng the effect on product colour and maximum wat er mark intens ity. It was veri fied
that the introduction of fine grain zirconi um silica te in the glaze composition is much
more effec tive for cons training the water mark. However, excessive addi tions can lower
glaze meltability unduly and raise its opacity, and can thus impede the application of the
decorati ve elements onto the glaze.
P. GI - 4+.'"
CASTELL6 :--: (SPAJ:--J)
Comnosition L* a* b*
Transparent Gla ze 89.7 0.52 4.00
Opaque Whit e Glaze 91.9 0. 13 1. 19
II QUALI
' .0 a.sa
as
ao
w
'."
1.5
<
1.0
0.5
0.0
White
Glazes
2002
Ta/1/t' Xll. Effect of the type of gla:c 011 product cotour Figllre 15. Effect of the type (:fsla:c 011 ,calt'/" mark
i lltt'l1:,ity .
Ccmnosition L* I a* b*
Transparent Glaze (ET) 89.7 I 0.51 4.00
ET +3.0% White fri t 89.4 0.51 4.08
ET + 5 ~ o Whit e fr it 89 9 0. 58 4.19
ET + 7.0% White fri t 89.8 0.54 4.15
ET + 3.0% Zirconium S. #500 90.4 0.5 1 4.00
ET + 5.0% Zirconium S. #500 90.5 0.43 3.67
ET + 7.0% Zirconium S. #500 91.0 0.36 3.63
3.'
3.2
3.0
E
2.'
,
E
2.6 ,

E
2A
W
<
2.2
2.0
i .e
.,
0
o
c
o
---0-- White fnt
-0- Zirconium silicate #500
6
Ad dition to the glaze (% )
Table XIll. Effect of white {rit and zi rconunn silicate
((l ll tt :ll f 011 I';lgorc cotour.
5. CONCLUSIONS
Figllre 16. EJf ecf ofichitcfrit and : ircol/illlll silicatt'
col/tent 011 the uvtcr mark:
Th e results obtained indicate that the water mark phenomenon ari ses preferentiall y
in high porosity ceramic tiles, and that the most effe ctive way of minimisi ng this is by
adjusting the engobe compositi on. The use of the col orimetric technique to quan tify the
water mark allows greater accuracy and sensitivity in the an alysis of the result s.
To avoid water mark formation it is fundamental for the engobe layer to p resent
high opacity. In this sense, the presence of hi gh zirconium silicate contents, or using
thicker engobe layers are measures that were shown to be effective. If the engobe layer is
excessively porous, opacity can be adversel y a ffected when the tile abso rbs water. For th is
it is fu ndamen tal for the engobe to have high meltabili ty, to provide the smallest possible
porosity in the product after firing.
The tests carried out indicate that frit , talc and fel dspar contents have a great effect
on engobe meltabilit y, while excessive zirconiu m silicate ad ditio ns are detr imental to the
meltability of the engobe, despite guaran teeing its opacity. The clays should also be
selected with due cri teri a, as they affec t engobe layer porosit y and hence can inc rease the
possibili ty of water marks appearing, as well as affecting colour.
Adjustments in the gl aze composition can also con tribute to mlt11mt Smg the
problem. The use of white zirconium glazes or small zirconium silicate additions to
tr anspar ent glazes also con tribu te minimisi ng the intensi ty of the water mark.
r.GI -l-4Y
II QUAUCW" 2002
6. REFERENCES
CASTELL6'J (SI'Al 'J )
(I] "Smal ti Cerarni ci" Grupo Ed itor iale Faenza Editrice S.p.A., 1992.
[2] BORASCHI, E.; CUNHA, LJ.V.; D. "Engobes: Carectertsticas e aplicacoes" Ceramica Ind ustrial, 01. 1, 31-33,
1996.
[3] PARMELEE, C. VV. "Ceramic Glazes" Third Edition, eBI Publishin g Company, USA. 1973.
[4J EPPLER, R A.; EPPLER, D.R. "Glazes an d Glas s Coatings", The American Ce ramic Society. CSA, 2000.
[51\VO:'\JG, C.W.; TA!'\, H.C.; L. !\,f. "Reverse staini ng . Problem of glazed ti les - Singapore Experience" wor kshop
on the latest International St and ards 7 Local Prob lems on Ceramic Tiles, Singa pore, 1995.
[6] !\l ELCHIADES, EG.; BOSCHl A. G. "Cor es e ton alidades em revest iment os ccra rnicos" Ce ramica Ind us trial, .l, (1-6),
11-18, 1999.
[7J MORE;....JO, A. ; BOU, E.; CABRERA, QCEREDA, P. " Zi rconi um silicate opacification me chani sm in cerami c
cngobcs" V World Congress on Cerami c TIle Qu al ity, Pus. 29-31, Castellon, Sp ain, 1998.
P. LI - -1 30