Journal of the European Ceramic Society 41 (2021) 5755–5767

Contents lists available at ScienceDirect

Journal of the European Ceramic Society

journal homepage: www.elsevier.com/locate/jeurceramsoc

Effect of strong mineral fluxes on sintering of porcelain stoneware tiles

Camila Tavares Brasileiro a, 1, Sonia Conte b, 1, Flávia Contartesi a, Fábio Gomes Melchiades c,
Chiara Zanelli b, *, Michele Dondi b, Anselmo Ortega Boschi a
a
Federal University of Sao Carlos, Graduate Program in Materials Science and Engineering, São Carlos, SP, Brazil
b
CNR-ISTEC, Istituto di Scienza e Tecnologia dei Materiali Ceramici, Faenza, Italy
c
Centro de Revestimentos Cerâmicos – CRC, São Carlos, SP, Brazil

A R T I C L E I N F O A B S T R A C T

Keywords: Strong fluxes are needed to fire vitrified ceramics at temperatures significantly lower than those usually reached
Ceramic tile in industrial firing cycles. This work is aimed at understanding the role of strong fluxes in the microstructural
Flux evolution during sintering. Six fluxes (colemanite, ulexite, wollastonite, diopside, spodumene and phonolite)
Melt composition
were individually added to a porcelain stoneware batch and processed in standard conditions. Compacts and
Porcelain stoneware
fired bodies were characterized by optical dilatometry, XRD-Rietveld, SEM and measuring technological prop­
Sintering
erties. Strong fluxes change the firing behaviour with a complex interplay of sintering kinetics, microstructural
features, and phase composition. Every flux has its own repercussion on the properties of the liquid phase
(chemical composition, degree of polymerization, viscosity and surface tension) which are key points to explain
the observed microstructure, densification rates, and stability at high temperature. Batches with phonolite,
wollastonite or diopside exhibit characteristics closer to standard porcelain stoneware, while spodumene and
borates suffer from unsatisfactory microstructures and lower densification efficiency.

1. Introduction 1210− 1240◦ C brings about a higher energy consumption and may
imply technological problems, such as compromising the refractories of
Ceramic raw materials are considered strong fluxes when they are the kiln [14,15] or promoting the pyroplastic deformation of tiles [8,10,
able to promote the viscous flow sintering at temperatures significantly 16]. In some cases, these constraints are particularly strict, e.g. when the
lower than those usual for more conventional fluxes, such as feldspar. national standards prescribe an even lower water absorption (<0.1% in
Examples of such fluxes encompass low melting minerals – like borates, Brazil) and available fluxes are not as strong as necessary, or the existing
zeolites and fluorine-bearing phases [1–4] – as well as feldspathoids industrial kilns cannot reach very high temperatures [3–5].
(nepheline, leucite, kalsilite) and several minerals of Li, Ca and/or Mg In the industrial practice, the addition of a strong flux to a given
[5–8] that do not necessarily exhibit a melting point below that of batch is designed empirically, typically by assessing its effect on tech­
feldspars. In addition, many silicate glasses perform as strong fluxes, nological parameters, like water absorption, linear shrinkage and me­
including scraps (e.g., soda-lime container or flat glass) and chanical strength. The presence of raw materials different from classic
glass-ceramic frits [9–12]. feldspathic fluxes, however, can affect sintering mechanisms and ki­
Strong fluxes are commonly used in the production of ceramic tiles as netics with a domino effect on other characteristics, like dimensional
sintering promoters to lower the firing temperature and, consequently, stability at the highest temperatures [3,10,11,16]. Thus, it is of para­
reduce the economic and environmental impact [2,3,13]. This is a key mount importance to comprehend how strong fluxes affect phase
target in the manufacturing of porcelain stoneware, where the attain­ transformations and particularly to which extent the amount, structure
ment of the very low water absorption prescribed by the product clas­ and properties of the melt formed at high temperature are changed
sification (<0.5 % according to ISO 13.006) often requires reaching the [17–19]. Some of the raw materials known for their powerful fluxing
highest temperatures within the usual firing range in the industrial characteristics are [3,20]: borates (ulexite and colemanite), diopside,
production of ceramic tiles (1180− 1240◦ C). As a matter of fact, firing at wollastonite, phonolite and spodumene. The use of these raw materials

* Corresponding author. Permanent address at: CNR-ISTEC, Via Granarolo 64, 48018, Faenza, Italy.
E-mail address: [email protected] (C. Zanelli).
1
Both authors equally contributed to the study.

https://doi.org/10.1016/j.jeurceramsoc.2021.04.027
Received 8 December 2020; Received in revised form 5 April 2021; Accepted 10 April 2021
Available online 22 April 2021
0955-2219/© 2021 Elsevier Ltd. All rights reserved.
C.T. Brasileiro et al. Journal of the European Ceramic Society 41 (2021) 5755–5767

in porcelain and porcelain stoneware bodies is not new and is shortly relevance of the viscosity of the liquid phase on viscous flow sintering
reviewed hereafter. and the differences in fluxing potential of the different fluxes used, the
Borates have gained a certain interest as fluxes in ceramic tile pro­ amount of each flux to be added was adjusted in accordance with the
duction, either as industrial minerals (ulexite, colemanite, borax, etc.) or predictions of the Fluegel’s model [45], as can be seen in Table 2. As a
as byproducts of boron mining and processing (see [4] and references consequence, the body compositions display a limited variation in silica
therein). Investigation about vitrified tiles has been mostly addressed to and alumina, with a simultaneous increasing of alkali and alkali-earths,
assess the technological feasibility of the introduction of borates depending on the flux. Despite the considerable variation of the flux
[21–28]. amounts, the chemical compositions of the batches are quite similar,
Diopside is employed by the ceramic industry essentially in Brazil except for the elements of interest. Details on the development of the
[29] and Russia [30–33], and the technological effects of its presence in formulations are given in a previous paper [46].
porcelain stoneware bodies were studied [34,35].
Wollastonite has been used mainly in the United States, India and 2.2. Processing
Russia in earthenware tiles bodies [36]. Elsewhere, it has been occa­
sionally used in vitrified bodies, as sintering promoter. In all cases the Each flux was ground to a residue similar to the standard body,
studies mainly focused on the effect of introducing wollastonite on the taking the coarser fraction (>45 μm, measured by wet sieving) as target
technological behaviour [37–39]. (between 3% and 8% by weight), then dry mixed with STD in a labo­
Nepheline phonolite, an extrusive equivalent of nepheline syenite, is ratory ball-mill for 30 s to admix without additional grinding. The batch
currently used in ceramic tile production in Brazil [40], Czech Republic, particle size distribution was determined by X-ray monitoring of gravity
France, Germany and Bulgaria [20]. There are no reports of proper sedimentation (ASTM C958). All curves are similar (Fig. 1) apart U4.0
studies about its effects. (slightly finer between 0.6 and 30 μm), while the absence of the curve
Spodumene has been used, in addition to its role as sintering promoter corresponding to C4.5 below 20 μm is due to slurry flocculation. The
[41], because of its effects on other characteristics, such as whiteness specific weight (SW) of both unfired and fired bodies was measured by
[42], mechanical properties [43], and control of pyroplasticity [14,44]. Helium pycnometry (ASTM C329).
The goal of the present study was to better understand the role of Powders were granulated (7% water) followed by uniaxial pressing
strong fluxes in the sintering of porcelain stoneware tiles. The effects of (Nannetti, Italy) of 6⋅2⋅0.5 cm tiles. The compaction pressure was
each flux on the sintering behaviour of a standard body was investigated adjusted to get a dry bulk density (BD) in the 1.93-1.95 g⋅cm− 3 range
focusing on microstructure, phase composition, and chemical-structural (geometrical determination, caliper with 0.01 mm precision). Total
features of the vitreous phase, in order to shed light on crucial aspects of porosity of powder compacts was calculated by the equation: (1 – BD/
ceramic tile production, such as sintering kinetics, efficiency of densi­ SW)*100.
fication and dimensional stability at high temperature. Tiles were fired in a laboratory roller kiln (Fortelab, Brazil) at
different temperatures with cycle of about 40 min (cold-to-cold). The
2. Experimental following technological properties were evaluated (ISO 10545-3): water
absorption, open porosity (OP) and bulk density (BD) ; total porosity by
2.1. Materials and formulations TP=(1–BD/SW)*100; closed porosity as CP = TP-OP; linear firing
shrinkage, as (Lm-Lf)/Lm*100, where Lm and Lf are the length of the
Strong fluxes were selected as commercial grade wollastonite, ulex­ mould and the fired tile, respectively (Table 3). The temperature
ite, colemanite, diopside, phonolite, and spodumene. They were chosen necessary to reach a water absorption <0.5 % (prescribed by the stan­
to obtain a broad range of chemical compositions of fluxes (Table 1). dard ISO 13006 for porcelain stoneware) was taken as the optimal firing
Two of them are borates: predominantly calcian and particularly rich in temperature (Tf) of each batch.
boron (colemanite) or soda-lime with half the amount of B2O3 (ulexite).
Further two are silicates with a significant amount of alkali-earths, such
2.3. Characterization of the fired samples
as CaO (wollastonite) or CaO and MgO (diopside). The last two are
mainly characterized by the presence of alkalis: Li2O in spodumene or
The phase composition was quantitatively assessed by X-ray powder
Na2O and K2O in phonolite (in amounts higher than conventional feld­
diffraction (Advance D8, Bruker, Germany) in the 10− 100◦ 2θ range,
spathic fluxes).
with a point detector equivalent time of 185 s per 0.02◦ 2θ scan step
The strategy adopted was to add a certain amount of each flux to a
(LynxEye, Bruker). A full profile modelling by Rietveld refinement was
standard porcelain stoneware body (STD). Taking into consideration the
performed with the GSAS-EXPGUI software package [47,48]. Mullite

Table 1
Chemical composition of the raw materials used as strong fluxes.
Ulexite (U) Colemanite (C) Wollastonite (W) Diopside (D) Phonolite (P) Spodumene (S)

SiO2 19.64 12.7 46.48 55.62 50.42 63.16

TiO2 0.69 – 0.01 0.04 0.33 0.05
B2O3 27.82 56.66 – – – –
Al2O3 6.84 2.12 0.37 3.1 20.9 27.45
Fe2O3 2.51 0.61 0.35 0.65 3.66 0.68
MgO 3.96 0.69 0.9 17.53 1.14 0.49
CaO 24.76 20.16 45.65 20.69 1.51 0.03
MnO – – – – 0.26 0.01
Li2O – – – – – 4.04
Na2O 12.01 3.7 0.02 0.54 10.4 0.18
K2O 1.52 0.48 0.08 1.02 4.33 0.14
P2O5 – – 0.1 0.06 0.07 0.04
SO3 0.17 – – – 0.23 –
Loss on ignition – 0.01 5.77 0.36 5.77 3.37

–: below detection limit. In addition, 0.11 % ZrO2, 0.07 % BaO, 0.11 % SrO in phonolite.

5756
C.T. Brasileiro et al. Journal of the European Ceramic Society 41 (2021) 5755–5767

Table 2
Calculated chemical composition of the batches.
% weight STD U4.0 C4.5 W6.5 D12 P14 S20
Strong flux added none 4.0% ulexite 4.5% colemanite 6.5% wollastonite 12% diopside 14% phonolite 20% spodumene

SiO2 71.36 69.29 68.72 69.74 69.47 68.43 69.72

TiO2 0.98 0.97 0.94 0.92 0.87 0.89 0.79
ZrO2 0.24 0.23 0.23 0.22 0.21 0.22 0.19
B2O3 – 1.11 2.55 – – – –
Al2O3 18.51 18.04 17.77 17.33 16.66 18.84 20.30
Fe2O3 2.19 2.20 2.12 2.07 2.01 2.40 1.89
MgO 2.34 2.40 2.27 2.25 4.16 2.17 1.97
CaO 0.17 1.15 1.07 3.13 2.63 0.36 0.14
Li2O – – – – – – 0.81
Na2O – 0.48 0.17 – 0.06 1.46 0.04
K2O 3.32 3.25 3.19 3.11 3.04 3.46 2.68
P2O5 0.04 0.04 0.04 0.04 0.04 0.04 0.04
Loss on ignition 0.11 0.11 0.11 0.48 0.14 0.90 0.76

–: below detection limit. In addition, 0.04 % MnO, 0.01 % BaO, 0.02 % SrO, 0.03 % SO3 in P14.

Fig. 1. Particle size distribution of the powders containing the various fluxes. The absence of the curve corresponding to C4.5 below 20 μm is due to slurry
flocculation.

Table 3
Physical and technological properties of batches (e.u.: experimental uncertainty).
Property unit STD U4.0 C4.5 W6.5 D12 P14 S20 e.u.

Median particle size μm 3.4 3.0 n.d. 3.7 3.7 3.5 3.7 0.1
Specific weight (unfired) g⋅cm− 3 2.706 2.660 2.658 2.678 2.695 2.632 2.707 0.003
Bulk density (unfired) g⋅cm− 3 1.95 1.94 1.93 1.94 1.95 1.94 1.95 0.01
Total porosity (unfired) %vol. 27.9 27.1 27.4 27.6 27.6 26.3 28.0 0.4
Firing temperature (Tf) ◦
C 1215 1140 1125 1185 1185 1170 1170 5
Specific weight at Tf g⋅cm− 3 2.557 2.549 2.547 2.545 2.531 2.544 2.529 0.003
Linear shrinkage at Tf cm⋅m− 1 7.9 6.6 4.9 7.7 8.0 7.9 7.6 0.2
Water absorption at Tf %wt. 0.1 0.4 0.2 0.1 0.1 0.3 0.3 0.1
Open porosity at Tf %vol. 0.28 0.97 0.46 0.22 0.27 0.80 0.70 0.1
Closed porosity at Tf %vol. 5.0 9.4 15.6 4.5 3.2 5.7 5.5 0.4
Bulk density at Tf g⋅cm− 3 2.43 2.31 2.15 2.43 2.45 2.40 2.39 0.01

stoichiometry was determined by the length of the unit cell a-axis, which weight of internal standard and Xs,c is the refined weight of the internal
scales linearly with the amount of Al2O3 [49]. The samples were standard.
admixed with 20% wt.Al2O3 as internal standard (Xs). The quantifica­ The chemical composition of the vitreous phase was estimated based
tion of the vitreous phase is based on the normalization equation ΣiXi ∕ = on the quantitative chemical and phase composition of fired body [18]
1, ΣiXi+Xa = 1, where Xi and Xa are the weight fractions of the crystalline by subtracting the contribution of each crystalline phase, considering its
components and the vitreous phase, respectively. Thus the amount of stoichiometric formula, from the bulk chemistry (the resulting values
the vitreous phase can be estimated directly from the weight of the in­ were normalized to 100 %). Such a vitreous phase contains different
ternal standard by: Xa= (100/80)[1-(Xs/Xs,c)] where Xs is the added elements that affect both structure and properties of the glass network.

5757
C.T. Brasileiro et al. Journal of the European Ceramic Society 41 (2021) 5755–5767

Table 4
Parameters used to express structural features of the melt (NBO/T, CCAT, GNF, GNM).
Parameter Definition

Degree of polymerization of the melt (NBO/T) From the composition of the liquid phase: number of Non-Bridging Oxygens (NBO) per Tetrahedrally-coordinated cations
(Si, B, Al) as atomic percentage
Charge compensators for aluminium in tetrahedral From the composition of the liquid phase: Al3+ charge compensated by alkali and alkaline earths: CCAT (atom%) = Na + K +
coordination (CCAT) 2Ca + 2Mg + Li (up to a maximum value = Al)
Glass network formers (GNF) From the composition of the liquid phase: GNF (atom%) = Si + B + CCAT, corresponding to Al3+ charge compensated by
alkali and alkaline earths
Glass network modifiers (GNM) From the composition of the liquid phase: alkali and alkaline earths exceeding CCAT: GNM (atom%) = Na + K + Mg + Ca +
Li – corresponding values in CCAT

In order to facilitate data interpretation, some parameters were used to the schedule went from Tf–20◦ C up to Tf by 10◦ C steps with heating
express specific structural features of the melt, as summarized in rate of 5 ◦ Cmin− 1 and 8 min dwell time at each step.
Table 4. 2) Sintering kinetics in isothermal conditions (ramp 80 ◦ C⋅min− 1 up to
The physical properties at high temperature were estimated by pre­ Tf and 30 min dwell time). Results were expressed in terms of
dictive models based on the chemical composition of the liquid phase. specimen volume as a function of dwell time. The relative density
The shear viscosity was calculated by the spreadsheet of Fluegel’s model was calculated by the starting bulk density and the volumetric
[45] because it includes specifically B and Li. It is worth noticing that shrinkage during firing. Different stages were identified with refer­
this model tends to underestimate viscosity of peraluminous melts [18]. ence to Fig. 2.
An alternative predictive method, proposed by Giordano and coworkers
[50], was used as a reference, having care of expressing B2O3 and Li2O as 3. Results
equivalent molar amounts of Al2O3 and Na2O, respectively. Surface
tension was estimated interpolating the data obtained by the Appen’s 3.1. Firing behaviour
[51] and Dietzel’s [52] methods.
The relative viscosity – i.e. the viscosity of the bulk (melt + crystals) The presence of a strong flux reduced the temperature necessary to
– was estimated by the volume fraction of solids (ϕs). At the maximum reach the water absorption prescribed for of porcelain stoneware (BIa
firing temperature, the rheological behaviour of porcelain stoneware group). However, this reduction varied considerably, in relation to the
bodies is similar to semi-diluted suspensions (0.25 < ϕs<0.40) pre­ reference body (Tf = 1215 ◦ C) as can be seen in Fig. 3.
senting a Newtonian behaviour. Thus, a simplified equation was used to The effectiveness of fluxes in reducing Tf is in the order: colemanite >
calculate the relative viscosity (dimensionless) as a function of the solid ulexite > spodumene ~ phonolite > wollastonite ~ diopside. This
content [19,53]: ηrel = (1–φs)− B, where B is the Einstein coefficient (2.5). ranking mirrors the interest of tile producers in the use of strong fluxes
The sintering behaviour was also evaluated by hot stage microscopy, [54,55], and matches the reported improved firing behaviour induced
using an optical thermo-dilatometer (TA, ODP868, Germany) which by colemanite [3,22] and diopside [29,34,56]. It is also in agreement
registered the size variation, determined by the pixel count, during the with the known behaviour of syenite and nepheline syenite, i.e. rocks
thermal cycle of specimens (5 × 5 × 5 mm in size) cut from the dry tiles. compositionally equivalent to phonolite [6,57,59,60]. Unexpectedly,
A volumetric shrinkage was calculated from the variation of the section the onset of the densification in the curve corresponding to the batch
area, assuming an isotropic behaviour during firing. Two targets were containing wollastonite was very similar to the STD. This apparently
pursued: contradicts previous observations [39,61].
In comparison with STD (Fig. 4A and Table 5), strong fluxes antici­
1) Firing behaviour in industrial-like conditions (heating ramp 40 pated the onset of densification: colemanite (− 88 ◦ C), ulexite (− 81 ◦ C),
◦
Cmin− 1) to evaluate the dimensional stability at high temperature, diopside (− 22 ◦ C), wollastonite (− 19 ◦ C) and phonolite (− 17 ◦ C). The
intensity of the ulexite effect may be overrated by the fact that body

Fig. 2. Schematic representation of an isothermal run with optical thermo-dilatometer of a typical porcelain stoneware body. Four different stages are indicated
along with the five turning points used to quantify the extent and rate of densification or bloating in each stage.

5758
C.T. Brasileiro et al. Journal of the European Ceramic Society 41 (2021) 5755–5767

(Fig. 2) was quantified (Fig. 5).

Stage 1, densification in the heating ramp. Only bodies containing bo­
rates have a densification equal or slightly higher than the STD
composition at this stage. The presence of wollastonite, phonolite,
spodumene and diopside clearly decreased the extent of densification
during the heating ramp, with the latter resulting in the greatest
decrease.
Stage 2, densification at constant rate. At this stage all the fluxes pre­
sented densifications larger than STD, with the exception of ulexite.
Diopside and spodumene produced the greatest increase, whereas
colemanite, wollastonite and phonolite exhibit intermediate effects.
Stage 3, densification at decreasing rate. Ulexite, wollastonite, diopside
and phonolite produced a small increase of the amount of densification
in comparison with the STD, whereas colemanite and spodumene pro­
duced a decrease.
Considering sintering curves in isothermal conditions (Fig. 6A) the
Fig. 3. Reduction of the optimal firing temperature (Tf) in relation to STD.
only curve similar to STD is D12. The final shrinkage scale with the bulk
density, even though differences are highlighted by the prolonged dwell
U4.0 has the finest particle size distribution of the set (Fig. 1 and time and a ramp (80 ◦ C⋅ min− 1) considerably faster than industrial
Table 3). schedules.
For all batches, densification proceeded initially at similar rates but Regarding the relative density curves (Fig. 6B), fluxes P14 and D12
important differences emerged in the final part of the curves in relation were capable of achieving values close to STD. In contrast, the maximum
to: relative densities achieved with the other fluxes were smaller, particu­
larly about S20C4.5 and U4.0. It deserves to be noticed that the plateau
(i) switch point – represents the point at which the curve changes its of stability at Tf (i.e., the time between the achievement of maximum
inclination, denoting a decrease of the densification rate; densification and the start of bloating) was rather short for D12 and
(ii) the degree of densification (i.e., the maximum volume variation); W6.5, whose volume shrinkage starts to decrease after 5− 7 min of dwell
(iii) the dimensional stability when both temperature and dwell time time. C4.5 has the shortest plateau (Fig. 6A).
increased.
3.2. Phase composition
Curves in Fig. 4B can be divided into two groups:
Phase transformations occurring during firing led to the composi­
• W6.5, D12, P14 and STD, with a delayed switch point and a high
tions summarized in Table 6. The introduction of strong fluxes induced a
degree of densification associated to a high dimensional stability, a
distinct phase assemblage with respect to the standard, according to the
slow bloating rate and a negligible density loss (Table 5);
chemical features of fluxes; moreover, it altered the chemical composi­
• C4.5, S20 and U4.0, where the switch point was anticipated and
tion and amount of liquid phase of the standard body (STD).
densification rate decreased faster than STD, resulting in a reduced
Two main trends can be observed: batches with borates or spodu­
degree of densification.
mene are characterized by contents of quartz and mullite comparable to
STD, leading to a slightly higher content of crystalline phases and, as a
The extent of densification by the previously illustrated stages
consequence, lesser vitreous phase, especially in sample S20, where

Fig. 4. Sintering curves of porcelain stoneware batches: specimen volume as a function of temperature (A) and time (B). Test carried out by hot stage microscopy,
using an optical thermo-dilatometer, with industrial-like schedules with prolonged dwell time and increased temperature to evaluate dimensional stability.

5759
C.T. Brasileiro et al. Journal of the European Ceramic Society 41 (2021) 5755–5767

Table 5
Firing behaviour of batches in the HSM tests (e.u. is the experimental uncertainty).
Property unit STD U4.0 C4.5 W6.5 D12 P14 S20 e.u.

Firing temperature, Tf ◦
C 1215 1140 1125 1185 1185 1170 1170 5
Start of densification ◦
C 971 890 883 952 949 954 972 2
Max densification (ΔVmax) %vol 24.5 17.4 19.4 22.1 24.0 22.9 19.3 0.1
Relative density at ΔVmax % 98.8 90.4 93.1 96.2 97.6 99.9 89.7 0.1
Early stage densification (<Tf) %ΔVmax 38.7 38.9 40.9 32.9 27.1 33.1 34.4 0.1
Constant rate stage (at Tf) %ΔVmax 23.5 19.5 25.8 27.1 33.7 27.6 34.0 0.1
Decreasing rate stage (at Tf) %ΔVmax 37.8 41.5 33.3 40.0 39.3 39.3 31.6 0.1
Relative density at switch point constant rate > decreasing rate % 86.2 80.3 84.7 84.4 85.0 87.8 81.8 0.1
Constant rate sintering at Tf min− 1 3.3 2.5 2.9 3.6 3.1 3.3 3.4 0.1
Decreasing rate sintering at Tf min− 1 0.18 0.17 0.36 0.36 0.33 0.33 0.29 .01
Density loss at Tf after prolonged firing time %vol <0.1 <0.1 1.0 0.3 0.2 <0.1 0.2 0.1

Fig. 5. Extent of densification stages: during the heating ramp (A); at constant rate at Tf (B); at decreasing rate at Tf (C).

Fig. 6. Sintering curves in isothermal conditions as volumetric shrinkage (A) or relative density (B) as a function of time.

some spodumene persisted in the final product. This was already of anorthitic plagioclase, which is documented in the literature on
observed in bodies with 7% colemanite [22] or 4–10 % spodumene [8, bodies containing either wollastonite [38,39] or diopside [29].
21,43]. Conversely, the batch with phonolite was more aggressive towards
On the other hand, batches with wollastonite, diopside or phonolite the crystalline phases, being the richest in alkalis. This peculiarity
showed a lower content of mullite with respect to the standard, while resulted in the highest content of vitreous phase, formed at the expense
quartz can be either diminished (W6.5, P14) or increased (D12). Among of quartz, mullite and feldspars. A similar behaviour was observed when
all samples, batches W6.5 and D12, particularly rich in alkali-earths, 15% nepheline syenite was added to porcelain stoneware [57]; at vari­
presented a low content of mullite and the highest content of feldspars ance, a slight decrease of mullite was found only when passing from 10%
among all samples. This was probably due to the formation during firing to 30% of nepheline syenite [58].

5760
C.T. Brasileiro et al. Journal of the European Ceramic Society 41 (2021) 5755–5767

Table 6
Phase composition of the body and chemical composition of the melt (e.u.= experimental uncertainty).
Parameter Unit STD U4 C4.5 W6.5 D12 P14 S20 e.u.

Tf ◦
C 1215 1140 1125 1185 1185 1170 1170 5
Mullite %wt 7.7 7.0 7.0 3.9 3.7 3.6 7.9 0.1
Plagioclase %wt 0.5 1.1 1.5 3.3 1.8 0.9 – 0.1
K-feldspar %wt – 0.7 0.8 1.4 1.0 – – 0.1
Diopside %wt – – – – 2.5 – – 0.1
Spodumene %wt – – – – – – 6.9 0.1
Vitreous phase %wt 64.6 61.9 63.1 65.6 62.1 71.1 57.5 0.5
Mullite: unit cell axis a nm .7570 .7570 .7561 .7570 .7570 .7570 .7570 .0001
Mullite: Al2O3 content %wt 75.3 75.3 74.3 75.3 75.3 75.3 75.3 0.5
SiO2 of the melt %wt 66.02 61.24 61.48 63.23 60.28 61.46 63.69 0.10
TiO2 %wt 1.53 1.58 1.50 1.42 1.41 1.27 1.39 0.01
ZrO2 %wt 0.37 0.37 0.37 0.34 0.34 0.31 0.34 0.01
B2O3 %wt – 1.81 4.08 – – – – 0.02
Al2O3 %wt 19.63 20.01 19.08 20.02 21.21 22.83 22.17 0.05
Fe2O3 %wt 3.42 3.58 3.39 3.19 3.26 3.43 3.33 0.03
MgO %wt 3.65 3.91 3.63 3.47 6.01 3.10 3.48 0.03
CaO %wt 0.11 1.51 1.23 3.81 2.64 0.46 0.25 0.02
Li2O %wt – – – – – – 0.47 0.02
Na2O %wt – 0.78 0.27 – 0.10 1.97 0.07 0.01
K2O %wt 5.18 5.10 4.89 4.44 4.66 4.95 4.73 0.03
P2O5 %wt 0.06 0.07 0.06 0.06 0.06 0.06 0.07 0.01
GNF %atom 39.76 42.96 43.47 41.46 41.97 40.14 38.25 0.20
GNM %atom 1.53 0.55 0.15 1.36 1.87 2.12 3.00 0.10
CCAT %atom 8.87 10.60 10.10 10.60 11.23 9.99 8.75 0.15
NBO/T %atom 0.22 0.14 0.14 0.12 0.10 0.23 0.25 0.01

In addition, 0.05 % MnO, 0.01 % BaO, 0.03 % SrO in P14; 0.01 % MnO in the other bodies.

Mullite composition was not affected by strong fluxes since the unit components [62]:
cell parameters remain constant (Table 6). The only exception was C4.5 - Glass Network Formers (GNF), which confer the polymeric
with a slightly lower alumina content, which is expected because Al2O3 structure to the melt, are Si, B and Al3+, this latter for the fraction
in mullite usually increases with the firing temperature. Indeed, as occurring in tetrahedral coordination.
observed by Conte et al. [19], the enrichment of alumina in the liquid - Charge compensators for aluminium in tetrahedral coordination
phase at high temperature reflects the mullite stability in contact with (CCAT), i.e. alkali and alkali-earths that compensate the Si4+–Al3+
the melt: the more mullite is fused with the increasing temperature, the charge mismatch and stabilize aluminium ions in four-fold oxygen
more enriched in alumina is the liquid phase. In turn, the residual (or coordination.
newly crystallized) mullite will exhibit a higher content of Al2O3, hence - Glass Network Modifiers (GNM), which cause the breakage of
a slightly bigger unit-cell volume. bonds between GNF and O, leading to a depolymerization of the melt
network. GNM are composed either by alkali and alkali-earths or by
3.3. Composition of the liquid phase Al3+. When the former are in excess with respect to aluminium, the melt
turns peralkaline. In contrast, when the latter is in excess of CCAT, the
The chemical composition of the liquid phase formed during the melt becomes peraluminous and Al3+ assumes an oxygen coordination
heating treatment depends on the starting bulk chemistry and the phase higher than four or give rise to Si-Al triclusters [62,63]. Usually, melts
transformations throughout firing (Table 6). The resulting melt formed in porcelain stoneware bodies are peraluminous, with a molar
composition can be better described in terms of its structural ratio between Al2O3 and alkali or alkali-earth oxides slightly exceeding

Fig. 7. Composition of the melt in terms of GNF, glass network formers–left; CCAT, charge compensators of tetrahedrally coordinated Al3+–middle; GNM, glass
network modifiers–right.

5761
C.T. Brasileiro et al. Journal of the European Ceramic Society 41 (2021) 5755–5767

the unit [17,19]. GNM is Mg.

These parameters of the various compositions are presented in The role of iron, titanium and zirconium dissolved in the melt is not
Table 6 and Fig. 7. The addition of strong fluxes did affect both amount discussed, even if their concentrations were not so low to be negligible.
and composition of GNF. As a consequence, the degree of melt poly­ This is due to their intermediate character that allows Ti4+ and iron
merization was different, as expressed by NBO/T (number of non- (which can occur as Fe3+ or Fe2+) to behave either as GNM or GNF in
bridging oxygens per tetrahedrally-coordinated cations). Three cases silico-aluminate melts [62]. Such a behaviour depends on factors that
stand out: bodies rich in alkali-earths (D12 and W6.5) had melts with Si cannot be predicted with the same reliability that is possible for alkalis
content similar to STD, but higher in Al (highly polymerized, NBO/T = and alkali-earths, and may have important repercussion on melt vis­
0.10− 0.12). Samples with borates (C4.5 and U4.0) presented melts cosity. Nonetheless, as the contents of Fe, Ti and Zr, were similar in all
richer in Si and Al (along with B, of course) with respect to STD (highly the melts under investigation, they are expected to have a similar in­
polymerized, NBO/T = 0.14). Batches rich in alkali (S20 and P14) fluence in all batches.
showed melts lower in Si and with variable Al content, resulting slightly
more depolymerized (NBO/T = 0.23− 0.25) than the standard (NBO/T 3.4. Physical properties of the liquid phase
= 0.22).
Since the bodies under study were rich in K and Mg, these elements The predicted physical properties of the liquid phase at Tf are shown
were always the major components of CCAT. On the other hand, the in Table 7. Melt viscosity, as calculated by the Fluegel’s spreadsheet
charge compensators vary according to the flux used, as the presence of [45], varies from 4.37 (STD) to 4.44–4.78 log10 Pa⋅s in the modified
Li in S20 or the relatively high amount of Na in P14. There is a hierar­ batches. Such higher viscosity values of the liquid phase, as a general
chical order in the charge-compensating cations, based on the different effect due to strong fluxes, is only apparent. Actually, it reflects pri­
affinity with AlO4 tetrahedra in the melt: K+>Na+>Ca2+>Li+>Mg2+ marily the lower firing temperature of the bodies bearing strong fluxes
[62]. This caused alkali and calcium to be fully engaged in the role of since shear viscosity is strongly dependent on temperature. Indeed, once
CCAT and explains why only Mg and Al behaved as GNM. Magnesium viscosity is calculated through the Fluegel’s model for every sample at
was the “main flux” (i.e., the actual network modifier beyond charge 1200 ◦ C, for sake of comparison, it emerges that strong fluxes reduced
compensators) in batches with borates, diopside or wollastonite. the melt viscosity with respect to STD (Table 7).
Conversely, in samples with phonolite and most of all spodumene – as The more pronounced effect takes place with borates, which are
well as in the standard body – it was Al3+ that behaved as GNM, because well-known to reduce the melt viscosity, particularly at relatively low
its content was too high to be fully balanced by CCAT. temperatures [62,65]. Literature reported a weaker effect of borate on
This is the first time that the melt composition in porcelain stoneware the shear viscosity, but regarding bodies with a mining waste [4]. In any
is considered in such a detail to explain the role of strong fluxes. Some case, boron can be incorporated in the tetrahedral framework by the
data exist only about batches containing nepheline syenite [57,58,64]. association of B3+ with charge-compensating cations. Hence, borates
Thus, speciation of Al3+ in the body is fundamental to understand have a depressing effect on the liquidus temperature, similarly to alkali,
how the element is partitioned between crystalline phases and the melt, but do not significantly affect the polymerization state of the melt [62].
where it can play the different roles of network former or modifier Boron may occur either in BO5− 3-
4 tetrahedra or in BO3 triangles; the
(Fig. 8). former units resemble the structural behaviour of Al3+ in aluminosili­
In all the batches, the majority of the Al3+ ions left the mineral cate melts [66–68]. This can explain the high degree of polymerization
responsible for its presence, and was incorporated into the melt, where of liquid phase in U4.0 and C4.5 bodies. Regarding the body containing
they behave mainly as GNF, or secondarily as GNM in STD, P14 and S20 spodumene, its moderate reduction of melt viscosity is not compatible
bodies. Aluminium retained in crystalline phases after firing was parti­ with the literature [21,43], however, the results reported in these papers
tioned among mullite (mostly in STD, U4.0, C4.5 and S20 batches) or to are based on indirect observations.
a lesser extent mullite and feldspars (in W6.5, D12 and P14). Moreover, In order to compare our data with the literature, viscosity was also
the high content of Al3+ in crystalline phases, observed for sample S20, estimated by the method of Giordano and co-workers [50], used in
was due to the content of spodumene (~7% wt), which acts as a buffer of recent overviews on porcelain stoneware tiles [18,19]. The viscosity
aluminium, too. calculated by this method (not reported in Table 7) is in the 4.2–5.0log10
The stoichiometry of the melts sees for borate-bearing batches (C4.5 Pa⋅s range, which is considerably lower than usually found in porcelain
and U4.0) the Al3+ all four-fold coordinated, acting as GNF and the stoneware bodies with potassium and mixed Na-K feldspars (5.0–5.3
liquid phase was poor in GNM (Fig. 7). On the other hand, the melts log10 Pa⋅s). A possible explanation for this difference is the relatively
present in S20, STD and P14 were slightly peraluminous, with Al3+ in high contents of iron oxide and alkali-earths in the batches under
excess to CCAT. Instead, the introduction of strong fluxes rich in alkali- investigation.
earths (W6.5 and D12) resulted in peralkaline melts, even though the Surface tension of melts varies from 0.347 N⋅m− 1 (STD) to
0.341− 0.369 N⋅m− 1 in the other batches. This difference primarily de­
scends from the lower firing temperatures allowed by strong fluxes. If
the surface tension is evaluated at 1200 ◦ C, the differences in relation to
the reference are reduced (Table 7). When compared with results re­
ported in the literature, the calculated values are higher than in K and
mixed Na-K porcelain stoneware, where surface tension is in the
0.33− 0.34 N⋅ m− 1 range [19].

3.5. Microstructure

The microstructure of the fired bodies can be seen in the SEM mi­
crographs of Fig. 9. The batches STD, P14, D12 and W6.5 are charac­
terized by a compact bulk structure with few pores, generally of small
dimensions (around 15 μm on average). The pore size increases in the
following order, STD, P14, D12, W6.5. These results are compatible with
previous studies that reported relatively big pores (10− 30 μm) formed
Fig. 8. Speciation of Aluminium in the different bodies. in wollastonite-bearing bodies [38,39]. The samples with lower degree

5762
C.T. Brasileiro et al. Journal of the European Ceramic Society 41 (2021) 5755–5767

Table 7
Physical properties of melts and bodies at high temperature.
Property STD U4 C4.5 W6.5 D12 P14 S20 e.u.

Tf ( C)
◦
1215 1140 1125 1185 1185 1170 1170 5
Melt viscositya at Tf (Log10 Pa⋅s) 4.37 4.67 4.78 4.44 4.53 4.53 4.60 0.02
Melt viscositya at 1200 ◦ C (Log10 Pa⋅s) 4.48 4.21 4.20 4.32 4.41 4.32 4.39 0.02
Melt surface tensionb at Tf (N⋅ m− 1) 0.347 0.357 0.349 0.341 0.369 0.359 0.361 .005
Melt surface tensionb at 1200 ◦ C (N⋅ m− 1) 0.349 0.359 0.349 0.339 0.367 0.355 0.357 0.02
Surface tension/viscosity ratio (μm⋅s− 1) 14.7 7.6 5.8 12.4 10.9 10.6 9.0 0.2
Fraction of crystalline phasesc φs (1) 0.354 0.381 0.369 0.344 0.379 0.289 0.425 .001
Relative viscosityd (1) 2.98 3.32 3.16 2.87 3.29 2.35 3.99 .01
a
viscosity calculated by the Fluegel’s spreadsheet.
b
melt surface tension calculated by interpolation of Appen’s and Dietzel’s models.
c
weight fraction (1 – vitreous phase) from Table 6.
d
relative viscosity calculated by the equation (1–φs)− B (see paragraph 2.3).

Fig. 9. SEM micrographs of bodies fired at Tf (scale bar =100 μm).

of densification (C4.5, U4.0 and S20) presented different microstruc­ witnessed by a low tendency to bloating and the negligible loss of
tures and pores morphology. In particular, S20 presented a bimodal density after prolonged time at Tf (Table 5).
distribution of porosity: relatively large pores (around 20− 30 μm in Such microstructures suggest different mechanical properties of the
diameter) together with a series of small, closely spaced pores, smaller obtained tiles. Generally, bending strength in ceramic tiles decreases
than 10 μm. Similar features were already observed in exponentially as porosity increases, and the denser the fired bodies the
spodumene-bearing bodies [21], where porosity depends on the higher the modulus of rupture. However, it does not only depend on the
moment at which crystallization occurred during firing [69]. percentage of porosity, but also on the pore size and shape. Thus, it is
Borate-bearing batches exhibit the less compact structure, charac­ known that spherical, uniformly distributed pores with a size in the
terized by the presence of relatively large pores, as already observed in 5− 20 μm range have a positive effect on bending strength [71], as
stoneware bodies containing colemanite [22] or ulexite [27]. This is expectable in the case of STD, D12, W6.5 and P14 batches. On the other
especially true in the case of sample C4.5, which has pores up to 150 μm. hand, the presence of big pores in S20 and mostly in the borate-bearing
U4.0 has the lowest relative density at the turning point between con­ batches can negatively affect the mechanical properties of the ceramic
stant rate and decreasing rate of sintering (Table 5), coupled with the body.
higher open porosity of the set (Table 3). It could seem that this batch
did not have the necessary conditions to close the open porosity and 4. Discussion
accomplish sintering, resulting in a poorly densified body. A possible
explanation for this is that it could be a case of anticipated over-firing The physical properties of the liquid phase determine the sintering
[70] even though the bulk density of unfired tiles is just slightly lower kinetics and also have a key role in the dimensional stability of the tile at
than STD. Despite this, it is quite stable at high temperature, as high temperature. Other important factors are also deformation

5763
C.T. Brasileiro et al. Journal of the European Ceramic Society 41 (2021) 5755–5767

phenomena and de-sintering processes [70,72]. It is necessary to bloating phenomena) and responsible of the density loss during dwell
consider both the melt and the ceramic body as a whole, with their time. However, the occurrence of coarser grains of flux cannot be
physical properties and microstructure. There is indeed an interplay of excluded, which is known to induce the formation of large pores [72].
stability at high temperature, efficiency of densification, microstructure, Anyway, the high relative viscosity, accounting for the contribution of
technological behaviour, and properties of the liquid phase. Taking into the skeleton, seems to prevent the sample C4.5 from collapsing.
account all these parameters, it is possible to point out general trends, Although the technological behaviour of these bodies substantially
together with specific issues, considering separately the batches with confirms what is known in the literature about the addition of ulexite
borates (U4.0 and C4.5), those with alkali-earth fluxes (W6.5 and D12), and colemanite to porcelain stoneware [3,22,27,73], these consider­
and the ones with alkali fluxes (P14 and S20). ations explain for the first time, through the role of boron in phase
A common feature to most batches with strong fluxes is the early equilibria and in the properties of the vitreous phase, the relationship
start of densification, which allows to obtain a water absorption below between microstructure, composition and sintering kinetics.
0.5 % at a lower firing temperature with respect to the reference
formulation. This advantage, however, can be at the expense of other
4.2. Wollastonite and diopside
characteristics, as it will be seen on a case by case basis.
Batches with wollastonite or diopside provided porcelain stoneware
4.1. Borates tiles with the same bulk density and water absorption of the reference,
although obtained with a lower firing temperature of 30 ◦ C. This was
Bodies with ulexite or colemanite are characterized by the worst possible thanks to an earlier start of densification, associated to a greater
efficiency of densification, with bulk density below 2.32 g⋅ cm− 3 and extension of the constant rate stage (in case of diopside addition) or a
relative density between 90% and 93% (Tables 3 and 5). These features faster sintering rate (when wollastonite was used). The switch in the
are reflected in a microstructure where large pores are present to a sintering curves between constant rate and decreasing rate occurred at
greater extent than in a typical porcelain stoneware (Fig. 9). The addi­ relative density values slightly lower than STD (Fig. 10). The densifi­
tion of a borate caused limited variations in the phase composition, cation efficiency is indeed a little lower, compared to STD. This is
resulting in a slight decrease in the quantity of vitreous phase, which is a confirmed by the microstructure, where the amount of large pores is
highly polymerized melt, where the fair boron amount still allows the slightly greater than the reference. These characteristics represent clues
viscosity to be kept relatively low. Since the firing temperatures of instability during dwell time, which are in agreement with a greater
(1125− 1140 ◦ C) are low for standard porcelain stoneware, the melt density loss at Tf (Table 5).
viscosity at Tf is even higher than that of STD. In these conditions, There is a significant change in phase composition, where the
however, the sintering rate was slower, as expectable by a lower surface reduction of mullite and the increase of feldspars are common features.
tension to viscosity ratio of the melt in the U4.0 and C4.5 batches. As a However, two distinct fates were found for the strong fluxes: wollas­
consequence of the rather high melt viscosity, a fair stability during tonite disappeared by melting (increasing the vitreous phase) or reacting
dwell time was registered. It was better for ulexite than colemanite, with quartz and clay relics to form anorthitic plagioclase; at variance,
because U4.0 has a higher relative viscosity with respect to C4.5 diopside was partially retained and its firing reactions apparently did
(Table 7). not involve quartz, resulting in the formation of lesser melt. Calcium
The causes of the observed reduction of sintering efficiency is related dissolved in the liquid phase acts as a charge compensator for Al3+,
with the switch between sintering stages (constant rate versus leading to a highly polymerized glass structure. This can explain why the
decreasing rate): the lower is the relative density at the turning point, melt viscosity remains quite high, despite the magnesium acts as the
the lower will be the efficiency of densification (Fig. 10). This reflects only glass network modifier.
the values of linear shrinkage and bulk density of C4.5 and U4.0, lower These observations represent a step forward compared to what
than the other bodies. C4.5 has a higher relative density at the switch known in the literature on the technological properties of porcelain
point and a lower water absorption at Tf with respect to U4.0. This stoneware containing diopside [34,35] or wollastonite [37,55] as they
means that big pores present in C4.5 are closed (likely formed by better explain the interrelationships between phase evolution,

Fig. 10. Efficiency of densification expressed as relative density (measured during isothermal run) at the maximum volume shrinkage versus the relative density at
the switch point between the sintering stages at constant rate and decreasing rate.

5764
C.T. Brasileiro et al. Journal of the European Ceramic Society 41 (2021) 5755–5767

microstructural characteristics, and sintering behaviour. microstructure, phase composition and sintering behaviour of
phonolite-containing bodies go beyond the current knowledge on por­
4.3. Spodumene celain stoneware and open a prospect for the successful use of fluxes
relatively rich in Fe2O3.
The addition of spodumene made it possible to fire 45◦ C less than the
reference, even though the resulting porcelain stoneware had slightly 5. Conclusions
lower bulk density and slightly higher water absorption. These differ­
ences are related to both the peculiar microstructure, with patches of Strong fluxes are able to significantly reduce the firing temperature
closely spaced small pores, and the worst efficiency of densification, of porcelain stoneware batches. On the other hand, they induce changes
with a final relative density close to 90 % (Table 5). This happens even if in phase composition, which reflect on the liquid phase in terms of
the sintering rates are faster than STD, since account must be taken that polymerization degree, composition of network modifiers, and Al
S20 starts densification at the same temperature as the reference and speciation. Such variations have relevant repercussions on both melt
that the switch between constant rate and decreasing rate in the sin­ viscosity and relative viscosity of the bulk. However, viscosity differ­
tering curves took place at a very low relative density (Fig. 10). These ences are damped by the temperature gap between the batches con­
results differ from the literature, where spodumene always tends to taining strong fluxes and the reference. The higher viscosity, due to the
improve densification [8,21,41–43]. The spodumene added in these lower firing temperature in presence of a strong flux, allows keeping
studies, however, is much less than the 20 % used in S20. This addition technological behaviour within the industrial practice. Nevertheless,
has apparently had small effects on phase equilibria, as quartz and effects on the efficiency of densification, microstructure and stability at
mullite are substantially unchanged. Nevertheless, there is a clear high temperature cannot be avoided.
decrease in the vitreous phase, compared to STD, since about one third The addition of borates allows a remarkable lowering of firing tem­
of the added spodumene is found at the end of firing. Although it is likely perature, through an earlier start of densification. At variance of typical
the product of dissolution-recrystallization reactions, we cannot exclude porcelain stoneware, sintering proceeds at a slower rate and with less
that some relict spodumene is preserved. efficiency, leading to a low bulk density. The liquid phase formed in
The liquid phase is less polymerized than that in STD, and has the these conditions is less and highly polymerized, resulting in a very low
greatest amount of glass network modifiers (Al3+). This entails a slight surface tension to viscosity ratio. Although the variations in phase
lowering of the melt viscosity [62,63]. However, the viscosity of the composition are limited, a detrimental effect on the microstructure is
liquid phase in S20 is higher than that in STD, due to the 45 ◦ C difference evident by the occurrence of relatively large pores, which seem con­
in firing temperature, and reflects on a clearly lower surface tension to nected with a poor stability during dwell time. The different behaviour
viscosity ratio (Table 7). of batches containing ulexite or colemanite depends primarily on their
The convergence of higher melt viscosity and solid content leads to a different content of B2O3.
significantly higher relative viscosity than the reference. This probably The addition of wollastonite or diopside promotes a reduction of
plays a role in the poor densification efficiency and porous microstruc­ 30◦ C in firing temperature for the same bulk density and water ab­
ture. It is also possible that spodumene dissolution-crystallization might sorption. This is achieved by an earlier start of densification, which
interfere with the densification process when the lithium addition is proceeds with sintering rates not far from usual in porcelain stoneware.
high [69]. Although the higher relative viscosity should be beneficial for The efficiency of the process, however, is somewhat lower than STD and
pyroplasticity, the density loss during dwell time was higher than the leads to a microstructure with a little more residual pores, which implies
lithium-free body, at variance of what found in the literature [8,44]. a certain instability during dwell time. This is the result of a change in
phase composition, with a reduction of mullite in favor of feldspars, and
4.4. Phonolite a consequent change in the amount and composition of the liquid phase.
Some differences emerge between wollastonite and diopside: the former
Phonolite is the least known ceramic flux among those studied here, reacts completely during firing, while the latter is partially preserved. In
although reference can be made to the literature on nepheline syenite [6, both cases, the melt is highly polymerized but the viscosity remains
57,58,60]. However, phonolite has a much higher iron content than close to that of the reference, due to the change in composition of charge
most commercial nepheline syenites. Phonolite made it possible to fire compensators and glass network modifiers.
45 ◦ C less than STD, but P14 had slightly higher water absorption and Alkaline fluxes allow a significant drop (-45◦ C) in firing temperature,
slightly lower bulk density. This is reflected by the microstructure, at the price of bulk density and water absorption values slightly worse
where the size distribution of skeletal grains and pores is substantially than the reference. The behaviour of phonolite and spodumene, how­
similar in P14 and STD. This goal was achieved thanks to the fact that ever, is clearly distinct. The addition of phonolite leads to a decrease of
densification starts earlier in the presence of phonolite and proceeds quartz and mullite compensated by an increase of the vitreous phase,
with the same sintering rate as STD, but with a longer duration of the which keeps composition and degree of polymerization close to those of
constant rate stage. This is clearly summarized by a higher densification the reference. The melt viscosity is higher in P14 than in STD, due to the
efficiency, with a delayed switch at the decreasing rate stage (Fig. 10). gap in firing temperature. This helps the body to keep stable during
The addition of phonolite induced a significant increase in the vit­ dwell time despite its lower solid content.
reous phase at the expense of quartz and mullite. Despite this, the liquid The body with 20% of spodumene has the worst efficiency of
phase maintains characteristics close to that in STD in terms of poly­ densification, despite the sintering rates are faster than the reference
merization degree, quantity and composition of the glass network batch. The switch between constant rate and decreasing rate of sintering
modifiers (aluminium). This circumstance led to a slight lowering of the is anticipated and leads to a microstructure with patches of closely
melt viscosity (once compared at the same temperature). However, at Tf spaced small pores. These results are in contrast with the literature,
the viscosity of the liquid phase is higher in P14 than in STD, due to the perhaps because of the large spodumene addition. Although quartz and
lower firing temperature. mullite are substantially unaffected, there is less vitreous phase because
The P14 body kept stable during dwell time, despite its relative some spodumene persists (unclear if residual or a dissolution-
viscosity is distinctly lower than the reference, as a consequence of the recrystallization product). The liquid phase is less polymerized than in
lower solid content. A possible explanation is in a different role of the Li-free body but, due to the lower firing temperature, has a higher
aluminium (and probably iron as well) in controlling the viscosity of the relative viscosity, as a consequence of greater solid content and melt
melt. This is probably an aspect that is not well considered by the viscosity. This can justify both microstructure and poor densification
method used in the assessment of viscosity. These observations on the efficiency, particularly if spodumene dissolution-crystallization actually

5765
C.T. Brasileiro et al. Journal of the European Ceramic Society 41 (2021) 5755–5767

interferes with the densification process. [25] E. Yersel, B. Tufan, T. Batar, A study of various boron additives on the mechanical
behaviour and microstructure of ceramic tiles, J. Ore Dress. 12 (24) (2010) 16.
[26] S.U. Bayca, T. Batar, E. Sayin, O. Solak, B. Kahraman, The influence of coal bottom
ash and tincal (boron mineral) additions on the physical and microstructures of
Declaration of Competing Interest ceramic bodies, J. Ceram. Process. Res. 9 (2) (2008) 118–122.
[27] S.U. Bayca, Effects of the addition of ulexite to the sintering behavior of a ceramic
body, J. Ceram. Process. Res. 10 (2) (2009) 162–166.
The authors report no declarations of interest. [28] S. Somany, G.G. Trivedi, T. Sridhar, A. Goel, D. Mohanty, B. Pitchumani, Increase
in vitrified tile production by the use of borate flux, Proceedings of QUALICER’14,
Acknowledgements Castellon, Spain (2014), 11 p.
[29] L. Kuzmickas, F.R.D. Andrade, G.A.J. Szabó, J.F.M. Motta, M. Cabral Jr, Influence
of diopside: feldspar ratio in ceramic reactions assessed by quantitative phase
This study was financed in part by the Coordenação de Aperfeiçoa­ analysis (X-ray diffraction-Rietveld method), Cermica 59 (350) (2013) 345–350.
mento de Pessoal de Nível Superior – Brasil (CAPES) - Finance Code 001. [30] V.I. Vereshchagin, L.Z. Reznitskii, E.P. Vasil’ev, Y.I. Alekseev, Diopside rocks—raw
materials for many purposes, Glass Ceram. 46 (1) (1989) 32–33.
[31] Y.I. Alekseev, V.I. Vereshchagin, E.A. Karpova, Influence of diopside on the
References formation of porcelain and its properties, Glass Ceram. 47 (9) (1990) 357–360.
[32] Y.I. Alekseev, E.A. Karpova, V.I. Vereshchagin, M.V. Ryabinina, P.I. Shatalov,
Phase formation and properties of electrical porcelain in the presence of diopside
[1] R. de’ Gennaro, P. Cappelletti, G. Cerri, M. de’Gennaro, M. Dondi, G. Guarini,
additive, Glass Ceram. 48 (7) (1991) 302–306.
A. Langella, D. Naimo, Influence of zeolites on the sintering and technological
[33] Y.I. Alekseev, A.E. Abakumov, E.V. Abakumova, Diopside porcelain, Glass Ceram.
properties of porcelain stoneware tiles, J. Eur. Ceram. Soc. 23 (13) (2003)
52 (4) (1995) 94–96.
2237–2245.
[34] A. Novaes de Oliveira, D. Hotza, M. Peterson, A. Oliveira da Silva, R. Piccoli,
[2] A.F. Gualtieri, Development of low-firing B-fluxed stoneware tiles, J. Am. Ceram.
Application of diopside mineral in low-temperature porcelain production,
Soc. 92 (11) (2009) 2571–2577.
Proceedings of QUALICER’02, Castellon, Spain, Castellon, Spain (2002) 169–172.
[3] F.G. Melchiades, L.R. dos Santos, S. Nastri, A.O. Boschi, Gres porcelánico
[35] M. Tarhan, Whiteness improvement of porcelain tiles incorporated with anorthite
esmaltado producido por vía seca: materias primas fundentes, Bol. Soc. Esp. Ceram.
and diopside phases, J. Therm. Anal. Calorim. 138 (2) (2019) 929–936.
Vidr 51 (2012) 133–138.
[36] G.M. Azarov, E.V. Maiorova, M.A. Oborina, A.V. Belyakov, Wollastonite raw
[4] C. Zanelli, E. Domínguez, C. Iglesias, S. Conte, C. Molinari, R. Soldati, G. Guarini,
materials and their applications (a review), Glass Ceram. 52 (9) (1995) 237–240.
M. Dondi, Recycling of residual boron muds into ceramic tiles, Boletín de la
[37] T. Manfredini, M. Romagnoli, M. Hanuskova, Wollastonite as sintering aid for
Sociedad Española de Cerámica y Vidrio 58 (2019) (2019) 199–210.
porcelain tile bodies, Int. Ceram. J. (2000) 61–67.
[5] V. Biasini, M. Dondi, G. Guarini, M. Raimondo, A. Argnani, S. Di Primio, Effect of
[38] S. Ke, X. Cheng, Y. Wang, Q. Wang, H. Wang, Dolomite, wollastonite and calcite as
talc and chlorite on sintering and technological behaviour of porcelain stoneware
different CaO sources in anorthite-based porcelain, Ceram. Int. 39 (5) (2013)
tiles, Silic. Ind. 68 (5) (2003) 67–73.
4953–4960.
[6] L. Esposito, A. Salem, A. Tucci, A. Gualtieri, S.H. Jazayeri, The use of nepheline-
[39] O. Turkmen, A. Kucuk, S. Akpinar, Effect of wollastonite addition on sintering of
syenite in a body mix for porcelain stoneware tiles, Ceram. Int. 31 (2) (2005)
hard porcelain, Ceram. Int. 41 (4) (2015) 5505–5512.
233–240.
[40] J.J. Aumond, L.F. Scheibe, O fonolito de Lages – SC, um novo fundente cerâmico
[7] M.F. Serra, M. Picicco, E. Moyas, G. Suárez, E.F. Aglietti, N.M. Rendtorff, Talc,
brasileiro, Cerã¢mica Ind. 1 (2) (1996) 17–21.
spodumene and calcium carbonate effect as secondary fluxes in triaxial ceramic
[41] J. Garcia-Ten, E. Sanchez, M. Tamlin, A. Sheth, O. Harder, J. Esteban, Use of
properties, Proced. Mater. Sci. 1 (2012) 397–402.
spodumene as a flux in porcelain tile compositions, Proceedings of Qualicer’00,
[8] T. Aydin, A. Kara, Effect of spodumene addition on pyroplastic deformation of
Castellon, Spain (2000) 139–141. Pos-.
porcelain stoneware, J. Ceram. Process. Res. 15 (6) (2014) 486–491.
[42] A. Gines, A. Orenga, A. Sheth, D. Thiery, Use of spodumene for manufacturing
[9] F. Matteucci, M. Dondi, G. Guarini, Effect of soda-lime glass on sintering and
porcelain tile bodies of high whiteness, Ind. Ceram. 996 (2004) 50–52.
technological properties of porcelain stoneware tiles, Ceram. Int. 28 (8) (2002)
[43] A. Tucci, L. Esposito, L. Malmusi, E. Rambaldi, New body mixes for porcelain
873–880.
stoneware tiles with improved mechanical characteristics, J. Eur. Ceram. Soc. 27
[10] E. Rambaldi, W.M. Carty, A. Tucci, L. Esposito, Using waste glass as a partial flux
(2-3) (2007) 1875–1881.
substitution and pyroplastic deformation of a porcelain stoneware tile body,
[44] D.Y. Hopa, E. Özel, Pyroplastic deformation behavior of vitreous porcelain bodies
Ceram. Int. 33 (5) (2007) 727–733.
in presence of spodumene, J. Aust. Ceram. Soc. 53 (2) (2017) 545–551.
[11] F. Andreola, L. Barbieri, E. Karamanova, I. Lancellotti, M. Pelino, Recycling of CRT
[45] A. Fluegel, Glass viscosity calculation based on a global statistical modelling
panel glass as fluxing agent in the porcelain stoneware tile production, Ceram. Int.
approach, Glass Technol. 48 (1) (2007) 13–30.
34 (5) (2008) 1289–1295.
[46] F. Contartesi, F.G. Melchiades, A.O. Boschi, Estimating the reduction of the firing
[12] C. Zanelli, G. Baldi, M. Dondi, G. Ercolani, G. Guarini, M. Raimondo, Glass–ceramic
temperature produced by the addition of energetic fluxes in porcelain tiles through
frits for porcelain stoneware bodies: effects on sintering, phase composition and
the Fluegel’s model, Cermica 65 (376) (2019) 605–610.
technological properties, Ceram. Int. 34 (3) (2008) 455–465.
[47] A.C. Larson, R.B. Von Dreele, GSAS, General Structure Analysis System, Los Alamos
[13] M. Dondi, M. Raimondo, C. Zanelli, Clays and bodies for ceramic tiles: reappraisal
National Laboratory, Los Alamos, NM, 1988, pp. 86–748.
and technological classification, Appl. Clay Sci. 96 (2014) 91–109.
[48] B.H. Toby, EXPGUI, a graphical user interface for GSAS, J. Appl. Crystallogr. 34
[14] C. Sadik, I.E. El Amrani, A. Albizane, Recent advances in silica-alumina refractory:
(2001) 210–213.
a review, J. Asian Ceram. Soc. 2 (2) (2014) 83–96.
[49] T. Ban, K. Okada, Structure refinement of mullite by the Rietveld method and a
[15] C. Zanelli, M. Dondi, M. Raimondo, G. Guarini, Phase composition of
new method for estimation of chemical composition, J. Am. Ceram. Soc. 75 (1)
alumina–mullite–zirconia refractory materials, J. Eur. Ceram. Soc. 30 (1) (2010)
(1992) 227–230.
29–35.
[50] D. Giordano, J.K. Russell, D.B. Dingwell, Viscosity of magmatic liquids: a model,
[16] L.R. dos Santos Conserva, F.G. Melchiades, S. Nastri, A.O. Boschi, M. Dondi,
Earth Planet. Sci. Lett. 271 (1-4) (2008) 123–134.
G. Guarini, M. Raimondo, C. Zanelli, Pyroplastic deformation of porcelain
[51] A.A. Appen, Chemistry of Glass (Russ.), Khimiya, Leningrad, Russia, 1974, p. 351.
stoneware tiles: wet vs. Dry processing, J. Eur. Ceram. Soc. 37 (1) (2017) 333–342.
[52] A. Dietzel, Relation between the surface tension and the structure of molten glass,
[17] C. Zanelli, M. Raimondo, G. Guarini, M. Dondi, The vitreous phase of porcelain
Kolloid-Z. 100 (1942) 368–380.
stoneware: composition, evolution during sintering and physical properties,
[53] D. Giordano, Advances in the rheology of natural multiphase silicate melts:
J. Non. Solids 357 (16-17) (2011) 3251–3260.
importance for magma transport and lava flow emplacement, Ann. Geophys. 62
[18] S. Conte, C. Zanelli, M. Ardit, G. Cruciani, M. Dondi, Predicting viscosity and
(2019) VO216.
surface tension at high temperature of porcelain stoneware bodies: a
[54] D.U. Tulyaganov, S. Agathopoulos, H.R. Fernandes, J.M.F. Ferreira, Influence of
methodological approach, Materials 11 (12) (2018) 2475.
lithium oxide as auxiliary flux on the properties of triaxial porcelain bodies, J. Eur.
[19] S. Conte, C. Zanelli, M. Ardit, G. Cruciani, M. Dondi, Phase evolution during
Ceram. Soc. 26 (2006) 1131–1139.
reactive sintering by viscous flow: disclosing the inner workings in porcelain
[55] L. Sánchez-Muñoz, J.B. Carda, E. Cerisuelo, F. Tuduri, S. Plaza, Elaboration of
stoneware firing, J. Eur. Ceram. Soc. 40 (2020) 1738–1752.
porcelain stoneware tiles at 1170◦ C by addition of wollastonite, Key Eng. Mater.
[20] M. Dondi, G. Guarini, S. Conte, C. Molinari, R. Soldati, C. Zanelli, Deposits,
206 (2002) 2205–2208.
composition and technological behavior of fluxes for ceramic tiles, Period.
[56] J. Wu, K. Li, X. Xu, Y. Zhang, X. Xu, X. Lao, White porcelain material based on
Mineralogia 88 (3) (2019) 235–257.
diopside, Int. J. Appl. Ceram. Technol. 14 (3) (2017) 454–460.
[21] J. Llop, M.D. Notari, I. Nebot-Díaz, E. Barrachina, I. Nuñez, J.B. Carda, Estudo da
[57] A. Salem, S.H. Jazayeri, E. Rastelli, G. Timellini, Dilatometeric study of shrinkage
gresificação de composições de grés porcelânico formuladas com matérias-primas
during sintering process for porcelain stoneware body in presence of nepheline
espanholas, Cerã¢mica Ind. 16 (5–6) (2011) 7–11.
syenite, J. Mater. Process. Technol. 209 (3) (2009) 1240–1246.
[22] N. Ediz, A. Yurdakul, Characterization of porcelain tile bodies with colemanite
[58] C. Zanelli, R. Soldati, S. Conte, G. Guarini, A.I. Ismail, M.S. El-Maghraby,
waste added as a new sintering agent, J. Ceram. Process. Res. 10 (4) (2009)
A. Cazzaniga, M. Dondi, Technological behavior of porcelain stoneware bodies
414–422.
with Egyptian syenites, Int. J. Appl. Ceram. Technol. 16 (2) (2019) 574–584.
[23] A. Moreno, J. Garcia-Ten, E. Bou, A. Gozalbo, J. Simon, S. Cook, M. Galindo, Using
[59] K. Kusumoto, T. Sugiyama, Development of low-temperature sintering stoneware
boron as an auxiliary flux in porcelain tile compositions, Proceedings of
bodies, Mater. Sci. Eng. 18 (22) (2011) 222030.
QUALICER’00, Castellon, Spain (2000) 77–91.
[60] E. Kamseu, T. Bakop, C. Djangang, U.C. Melo, M. Hanuskova, C. Leonelli, Porcelain
[24] S.M. De Paula, A. Albers, Effect of borates on the vitrification behaviour of a
stoneware with pegmatite and nepheline syenite solid solutions: pore size
porcelain type tile body, Proceedings of QUALICER’02, Castellon, Spain (2002).

5766
C.T. Brasileiro et al. Journal of the European Ceramic Society 41 (2021) 5755–5767

distribution and descriptive microstructure, J. Eur. Ceram. Soc. 33 (2013) [68] B.C. Schmidt, N. Zotov, R. Dupree, Structural implications of water and boron
2775–2784. dissolution in albite glass, J. Non. Solids 337 (3) (2004) 207–219.
[61] S.R. Bragança, H.C.M. Lengler, C.P. Bergmann, Wollastonite as a flux for ceramics [69] P. Descamps, J. Tirlocq, M. Deletter, F. Cambier, Composite approach for the
bodies, Mater. Sci. Forum 727 (2012) 1016–1021. reinforcement of ceramic floor tiles, Proceedings of QUALICER’00, Castellon, Spain
[62] B.O. Mysen, P. Richet, Silicate Glasses and Melts, 2nd ed, Elsevier, Amsterdam, (2000) 379–394.
2018, 708 p. [70] F. Contartesi, F.G. Melchiades, A.O. Boschi, Anticipated Overfiring in Porcelain
[63] C. Le Losq, D.R. Neuville, P. Florian, G.S. Henderson, D. Massiot, The role of Al3+ Tiles: effects of the firing cycle and green bulk density, Boletín de la Sociedad
on rheology and structural changes in sodium silicate and aluminosilicate glasses Española de Cerámica y Vidrio 58 (2) (2019) 69–76.
and melts, Geochim. Cosmochim. Acta 126 (2014) 495–517. [71] M. Romero, J.M. Pérez, Relation between the microstructure and technological
[64] A. Salem, S.H. Jazayeri, E. Rastelli, G. Timellini, Kinetic model for isothermal properties of porcelain stoneware. A review, Mater. Construcción 65 (320) (2015)
sintering of porcelain stoneware body in presence of nepheline syenite, e065.
Thermochim. Acta 503 (2010) 1–7. [72] H.J. Alves, F.G. Melchiades, A.O. Boschi, Effect of spray-dried powder
[65] D.B. Dingwell, R. Knoche, S.L. Webb, M. Pichavant, The effect of B2O3 on the granulometry on the porous microstructure of polished porcelain tile, J. Eur.
viscosity of haplogranitic liquids, Am. Mineral. 77 (5-6) (1992) 457–461. Ceram. Soc. 30 (6) (2010) 1259–1265.
[66] M.E. Fleet, S. Muthupari, Coordination of boron in alkali borosilicate glasses using [73] F.G. Melchiades, L.R. dos Santos, S. Nastri, A.O. Boschi, Alternatives in the body
XANES, J. Non. Solids 255 (2-3) (1999) 233–241. composition of porcelain tiles produced by the dry route, CFI-Ceram. Forum Int. 89
[67] L.S. Du, J.F. Stebbins, Nature of silicon− boron mixing in sodium borosilicate (1) (2012) E39–E43.
glasses: a high-resolution 11B and 17O NMR study, J. Phys. Chem. B 107 (37)
(2003) 10063–10076.

5767