Li2OSiO2Al2O3MeIIO Glass-Ceramic Systems for Tile Glaze Applications

I

~

,

?L%

J Am cerarn

soc 74 151983-87

(1991)

Li20-Si02-AIn03-Me"0Glass-Ceramic Systems for Tile Glaze Applications Cristina Leonelli, Tiziano Manfredini, Mariano Paganelli, and Gian Carlo Pellacani* Dipartimento di Chimica, University of Modena, 41 100 Modena, Italy

Jose Luis Amoros Albaro, Jose Emilio Enrique Navarro,* and Maria Jose Orts lnstituto de Quimica Tecnica (Tecnologia Ceramica], Universidad de Valencia, Asociacion de lnvestigacion de las lndustrias Ceramicas (A.I.C.E.),12004 Castellon, Spain

Silvia Bruni and Franco Cariati Dipartimento di Chimica lnorganica e Metallorganica, University of Milan, 20133 Milan, Italy

glasses. The objective was to determine the physicochemical properties, which would be a useful reference and guideline for selecting a candidate as a tile glaze component.

In order to verify the possibility of using glass-ceramic materials as tile coatings, the devitrification processes of three industrial formulations belonging to the Li2O-AI2O3-SiO2 glass-ceramic system were investigated by differential thermal analysis, X-ray diffractometry, scanning electron microscopy, and IR spectroscopy. Compositional variations were made by addition of large amounts of MgO or CaO or PbO (ZnO) oxides as well as through smaller additions of other oxides. In these systems the surface crystallization contributes appreciably to the bulk crystallization mechanism. All the systems investigated show a high tendency toward crystallization even at very high heating rates, developing a very close network of interlocked crystals of synthetic fl-spodumene-silica solid solutions (LiA1Si4010).The results of this research are expected to establish the conditions under which these glass-ceramic systems can be practically used as tile glazes. [Key words: glass-ceramics, tile, glazes, devitrification, additives.]

11. Background

The outstanding properties of the materials based on the Li20-Al2O3-SiO2 glass-ceramics are the low coefficient of linear thermal expansion, the high thermal shock resistance, and good chemical resistance, which determine a wide variety of application^?-^ In order to design a glass-ceramic material with the desired properties, it is necessary to achieve a thorough knowledge of the devitrification kinetics of the LizOAI2O3-SiO2 system, including a thorough understanding of the roles played by nucleating oxides.7-" This enables the optimization of the heat treatment by the addition of nucleating agents. The use of different nucleating agents, heating cycles, and temperatures can produce transparent glass-ceramics in which the main crystalline phase is a high quartz solid solution' or opaque glass-ceramics, in which the main crystalline phase is a keatite solid s ~ l u t i o n . ~ In particular, controlled crystallization of stuffed p-quartz (with substituting cat ions) or P-spodumene (Li20-A1 2 0 3 4Si02 to Li20-A1z03-10Si02) solid solutions results in glassceramics with near zero thermal expansion. This is because these systems present small negative and positive volume coefficients of thermal expansion, respectively, alon with minor crystalline phases and a residual glassy matrix.' l3 Attempts to accurately identify the phases which are present sometimes failed, resulting in unknown phases as reported in previous studies. l2 A variety of sinterable p-spodumene glass-ceramics have been achieved with a wide range of proper tie^.'^ Selection of an appropriate composition should be based on desired properties. The glass composition of 1:1:6 (lithia-alumina-silica) stoichiometry, with alumina excess, within the P-spodumene primary field in the Li20-A1z03-SiOz ternary phase diagram,'5,'6 closely matches commercial formulation^'^ and has recently been closely investigated by comparing the effects of the more established additives, TiOz and Ta205,on the devitrification of these glasses and the transformation process of intermediate phases." Literature data on current knowledge of P-spodumene structure, the mechanism for nucleation, and of the phase separations of lithia-alumina-silica glasses are reported in Refs. 14 and 18.

I. Introduction

I

ceramic tile industry, the change in firing technology from the traditional double firing to the fast single firing process has involved different formulations and compositions of the glazes, based on new types of glasses.' To date, glassceramic systems have never been used as glaze components. Their practical utilization was not feasible with the current firing technologies, because they require a controlled nucleation and crystallization process at increasing temperature in a relatively short time and a sintering process in presence of a liquid phase with formation of a very compact microcrystalline structure? Recently some industrial formulations, based on the classical Li2O-AI2O3-SiO2 system, have found a practical application as tile glazes, resulting in finished ceramic floor tile products having enhanced abrasion and wear resistance with improved mechanical properties. In order to understand the evolution of these unique properties, this paper is aimed at investigating the nonisothermal devitrification, the crystallization mechanism, and the effect of the glass matrix on the development of the crystalline phase of three multicomponent Liz0-Al2O3-SiO2-Me'*O (Me = Mg, Ca, and Pb) N THE

G. H. Beall-contributing

!

editor

111. Experimental Procedure Manuscript No. 197454. Received June 26, 1990; approved January 3,

Industrial grade raw materials were used for all the formulations prepared. The compositions studied derived from

1991.

*Member, American Ceramic Society.

983

Vol. 74, No. 5

Journal of the American Ceramic Society -Leonelli et al.

984

industrial formulations belonging to the classical LizOAI2O3-SiO2glass-ceramic system and are reported in Table I. Glasses were prepared by melting the raw materials in a continuous industrial smelting tank for approximately 2 h at 1400°C to homogenize the melt, and then were quenched in water. After drying, a portion of the glass was ground in an alumina ball mill (300 mL). The frit was then sieved through 100, 120, 230, and 325 mesh sieves, allowing three different particle size samples: lower than 45 pm, between 45 and 63 pm, and between 125 and 150 pm. One more sample with particle size lower than 20 pm was obtained by wet milling for 2 h the remaining glass in distilled water. These were used for thermal investigations. The quenched glasses were cut for SEM* sample preparation and were also powdered for X-ray diffraction and IR spectroscopic characterization. Characteristic glass transition temperature (T,) and crystallization temperature (T,) were determined using differential thermal analysis (DTA).+The heat treatment was carried out on about 50 mg of sample in a Pt crucible (0.08 mL of capacity) in the apparatus against the same amount of fired kaolin as reference. Static nonisothermal experiments were performed by heating glass samples from 25" to 1000°C at the rates of 5, 10, 20, and 50"C/min. Data for all the runs were collected directly from the DTA. To confirm the noncrystalline structure of the parent glasses and to detect and identify the crystalline phases formed during the heat treatments, X-ray diffraction was performed. Appropriate powders of quenched glasses and of glass-ceramics, obtained by firing quenched glasses at 900°C for 2 h in a Pt crucible, were investigated by X-ray (Ni-filtered CuKa powder diffractometer (XRD)); with a 2-s time constant and 1000-count rate in the 28 range from 10" to 40" at a scanning speed of l"/min. All the systems were also investigated by infrared spectroscopy. The transmission IR spectrag of powdered glasses and glass-ceramics were recorded on KBr pellets in the range 400 to 4000 cm-'. To verify the devitrification capability of these systems at very fast firing rates, which are used in the modern tile industry, thin (about 1-mm) films of quenched glass coatings on a clay support of about 2 cm2 were heated at 1060°C for 5 min to simulate fast-firing industrial cycles. XRD measurements were directly performed on the thin film coatings, and the specular reflection IR spectras of the coating films were recorded at nearly normal incidence (12") in the range 400 to 4000 cm-'. SEM* measurements were also performed on these systems for microstructural characterization of the developed crystalline phase.

IV. Results and Discussion Three compositions (Table I) were melted into glasses which were optically clear, bubble free, and homogeneous to

*Model PSEM 500, Philips, Eindhoven, Netherlands. +STA409 thermobalance, Netzsch, Selb, FRG, interfaced with 9000/300 computer, Hewlett-Packard, Palo Alto, CA. 'PW MOD 1050, Philips. 'FTS-40, Digilab, Cambridge, MA.

Table I. Cornnosition of Glasses Investigated Cornoonent

MEO elass

Si02 TiOz A1701 Li;OPbO MgO ZnO CaO

60 5 6 14.5 13

Content (mol%) PbO elass

65 6 6 8.5 6 7

CaO elass

60 6 17.5

15

polarized light. All compositions also contain the nucleating P205oxide and other minor oxides (1% to 1.5%), specifically chosen to lend the bulk glass-ceramic materials the required properties. Furthermore, in the MgO system the presence of Ti02 promotes the known twin effects of developing the nucleating phase MgTiz05 and entering in the metastable and stable lithium aluminosilicate stuffed derivatives (&quartz and ~-sp~dumene(ss)).~~~~-~~ In the PbO glass, variables of interest are the presence of PbO and ZnO. PbO, when substituted for a similar molar amount of L i 2 0 , decreases the crystallization rate of the glass system. ZnO, in substitution of a similar molar amount of MgO, controls the bulk expansion coefficient of the residual glassy phase and in the presence of small amounts of P205oxide also seems to play a role as a nucleating agent." Coefficients of thermal expansion of the glasses and of the bulk crystallized materials (Table 11) must be strictly controlled, since they must be compatible with the clay bodies used in the tile industry, which generally present CTE values ranging from 6.5 x to 7.5 x low6. The DTA thermograms for the three glass formulations investigated are shown in Fig. 1. Three very similar features of each thermogram are apparent: (a) An endothermic event, corresponding to an increase of specific heat, signifying the glass transition (7''). It is worth noting that all the compositions investigated presented very similar Tgvalues obtained at different runs performed in the same experimental conditions. Since it is known that at the glass transition temperature all the glassy systems present very similar viscosity values to 1013.5Pa), it can be hypothesized that, for the investigated systems, the changes in the bulk atomic structure are somewhat similar despite the difference in cation composit i o n ~(b) . ~ An ~ exothermic event indicating crystallization (T,) with a maximum at a temperature of about 732", 711", and 746°C for the MgO, CaO, and PbO glasses, respectively. This was lower than that found for certain P-spodumene glasses (871"C),14 but confirms the temperature ranges and values previously observed in glasses of similar comp~sition.'~ Furthermore, as suggested by the peak sharpnesses and heights, the crystallization rate decreases in the order MgO > CaO > PbO glasses, consistent with the presence of TiOz and with the amount of L i 2 0 oxides, even though the CaO glass exhibits the lowest crystallization onset. (c) An additional endothermic event, occurring just after the crystallization completion, probably involving the melting of an undefined phase. It is known that, in glass-forming systems, surface crystal nucleation generally occurs more easily than internal crystal nucleation, which generally requires the addition of nucleating agents. In order to evaluate the bulk nucleation properties of these systems, samples with different specific surface area were studied by DTA. The increases of the specific surface area did not significantly influence the glass transition temperatures, whereas the crystallization peak maxima shift toward lower temperatures (Table 111). This behavior indicates that in the course of bulk crystallization, even though all these systems are internally nucleated (P205,Ti02),surface crystallization plays a significant greater in the PbOcontaining glass than in the others. In Fig. 2, the shift of the crystallization peaks due to the different particle size distributions is shown only for the MgO system, since the behavior of the other investigated systems is very similar. Table 11. Thermal Coefficient of Expansion of the Glasses and of the Bulk Crystallized Materials CTE ( x 106/OC)* , ~, Bulk crystallized material ~

System

Glass

MgO glass CaO glass PbO glass

7.32 8.57 6.96

*Measured from 25" to 300°C.

4.89 6.07 6.60

May 1991

LizO-Si02-Al~03-Me"0Glass-Ceramic Systems for Tile Glaze Applications

985

Temperature ("C) Fig. 1. DTA thermograms of the (a) MgO, (b) CaO, and (c) PbO glass system (heating rate 20"C/min).

Table 111. Maximum Peak Temperatures ("C) for the Glass Systems at Different Heating Rates and Granulometries Glass svstem

5

MgO glass CaO glass PbO glass

703 668 700

Heating rate ("C/min)* 10 20 50

723 682 722

732 711 746

760 747 792