During firing, a glaze applied on a ceramic body undergoes several transformations, including water loss from the clay components, glass transition and softening. During the softening phase, the glaze starts to melt, giving rise to a continuous liquid layer. As it cools, the glaze viscosity increases until the glaze becomes rigid and starts to contract simultaneously with the ceramic body (though not necessarily in the same way). The temperature required to achieve this locking phenomenon between the two materials during cooling is called the coupling temperature. Corresponding to the coupling temperature, the glaze softens during heating (absorbing tension) and solidifies during cooling (building up tension).

In glazed or double-layer tile, deformations may be generated from the different behavior of the two overlapped layers during both the heating and cooling phases. The coupling of materials with different thermal behaviors inevitably gives rise to a system of stresses due to the thermal incompatibility between the layers.

Traditional Issues with CTE Evaluation

The traditional method of studying differences in the coefficient of thermal expansion (CTE) involves looking at the single thermal expansion curves of a body and glaze without taking into account that, once “coupled,” they constitute a unique mechanical system. This approach is mainly due to the analytical limits of traditional dilatometers in the viscous-elastic region of the two layers, when tensions are primarily established.

Tensions due to thermal expansion mismatch in a bilayer system can result in: bending, up to the extent allowed by the materials’ mechanical strength; or cracking, when the fracture threshold is overcome. However, it is a common practice in the ceramic tile industry to use differential cooling on the two surfaces of a tile to control the planarity of the finished product. That is, depending on concavity or convexity of the already-fired tile, the glaze surface is cooled faster or slower than the substrate.

CTE mismatch and interphase reactions are the key parameters that need to be controlled during the firing cycle. Potential problems and solutions include:

• CTE differences lead to a concave tile; a faster cooling cycle applied to the side with glaze induces the glaze to develop a state of compression, hence the tile will be flat.
• CTE differences lead to a convex tile; a slower cooling cycle is applied to the side with the glaze, developing a state of tension and leaving the tile flat.

This completely empirical and unmethodical approach relies solely on the kiln technician’s visual evaluation and results in a certain amount of waste tile. In addition, this experimental approach is far from reproducible and needs to be replicated through trial and error every time a different tile is produced, since body and glaze CTEs change according to their chemical compositions. This kind of post-correction also inexorably gives rise to a system of tensions that are needed to keep the system planar, thus overcoming the effects of the CTE differences. This state changes from batch to batch and cannot be characterized by any dilatometer, regardless of design.

Modern Processes

The ceramic tile manufacturing process has evolved in multiple ways. Fast firing is now the dominant technology for tile manufacturing, with the time available to reach optimal body stabilization being reduced to just a few minutes. Tile sizes are also growing quickly; the market now demands tile of 1 m (3.3 ft) and larger. At the same time, tile thickness has been reduced to the minimum in order to lower transportation and other costs.

The value of the product lies in the richness of the surface, so glaze thicknesses are increasing. Geometrical perfection is another must: tile are mechanically squared to give a marble-like look to the surface. Therefore, the major challenge now is planarity.

If we consider a porous wall tile, the old trick of increasing the state of compression of the glaze onto the body to avoid delayed crazing used to work well on smaller, thicker tile. However, the same level of compression on the very same body and glaze—but with a much larger size—has a nasty side effect: bending. The only way to alleviate bending is to minimize the amount of compression. A second defect, peeling, could result if a glaze’s compressive tension is too high. In some cases, the tendency of the glaze to peel off the body appears in the sharp corner of three-dimensional products, such as kitchenware.

To avoid defects, it is important to determine the conditions yielding the minimum amount of compression. Since all glazes, like glasses, are characterized by a very low resistance to traction stress, a small amount of traction may cause their rupture. If the body expands even by a small percentage (e.g., because of moisture absorption), the glaze may easily crack unless it is in a state of compression with respect to the body. In this case, a little expansion of the body reduces the state of compression without generating a dangerous traction. This tendency to react with water in the course of time cannot be completely eliminated; all glazed bodies show this problem. (The only type of body that can be considered stable during this time is a completely sintered body, characterized by a lack of porosity.) The problem, however, can be addressed and resolved by using thermal expansion and bending tests to study the state of tension at the interface between glaze and body.

In terms of theoretical deformations, coupling between glazes and ceramic bodies was studied by Amoros, Negre, Belda and Sanchez, who used a formula derived from the Timoshenko equation to theoretically calculate the curvature induced in a glazed ceramic tile by the state of tension (compression/traction) of the glaze. They introduced some simplifying hypotheses, including isotropic, homogeneous, perfectly elastic materials; no interface development between the support and glaze; using the same temperature among the layers, and an elasticity moduli ratio constant during cooling (see Figure 1):

where D = deformation intended as “deflection” (mm), L = length (mm), h = thickness (mm), ∆c = percentage difference between the single dilatometric curves of the ceramic body and the glaze at room temperature (after translating the glaze’s dilatometric curve so that it coincides with the body’s dilatometric curve in correspondence of the coupling temperature), and

where S_s = support thickness, S_g = glaze thickness, E_s = support elasticity modulus, and E_g = glaze elasticity modulus. (It should be pointed out that the ∆c value needs an experimental determination.)

In a further publication, Amoros and Moreno clarified that, during firing, diffusion and dissolution phenomena between the glaze and support layers occur and a glaze-body interface develops. Thus, glaze and body do not behave as independent layers because of the presence of the interface, which affects the properties of the final product, including planarity. For this reason, they state that coupling between glaze and support should be investigated experimentally, using experimental conditions that better reproduce industrial settings.

Experimental Method

Considering the limits of determining theoretically the flexion behavior of a glazed ceramic material, experimental methods are of fundamental importance for studying the deformations induced by the state of tension established between the glaze and ceramic support. Traditionally, the instrument used to measure coupling temperature was a tensiometer. This was problematic, as the test was performed with a very long (30 cm) specimen that was not placed completely inside the kiln. The result was a significant lack of homogeneity for the sample temperature that resulted in unreliable data.

Optical techniques enable the characterization of the material’s behavior during firing and cooling without entering in contact with the specimen (with no interference caused by the measuring system), providing a good understanding of the material’s behavior in an actual industrial firing cycle. An optical fleximeter was used to study the deformations and state of tension in ceramic materials. During testing, a small sample bar is suspended between two holding rods spaced at 70 mm while a camera frames the center of the sample, which moves downward or upward during the heat treatment (see Figure 2). The beam of blue light that lights the center of the specimen has a wavelength of 478 nanometers and enables the digital camera to reach an optical resolution of 0.5 micron/pixel. The curve obtained with this instrument allows us to identify the coupling temperature between glaze and ceramic body, to obtain information about the sample planarity and pyroplastic behavior, and to qualitatively study the state of tension established between glaze and body.

Results and Discussion

As illustrated in Figure 3, a two-way heating cycle (20°C/min heating rate and slow cooling) was applied on a glazed single-fire (monoporosa) sample. Absolute bending values remain within 100 microns because, up to the glaze melting (during heating) and from the coupling temperature (during cooling), bending occurs mainly as a result of a difference in thermal expansion (i.e., elastic deformation) between the layers. The downward flexion in the initial part of the curve is due to the differences between the coefficients of thermal expansion of the glaze and body; since the body has a higher CTE compared to the glaze, it is subjected to a higher expansion, becoming longer than the glaze.

The specimen appears concave. The curve shows a negative peak at 600°C, after the glass transition (a quartz to b quartz) occurring in the ceramic body. The difference between the thermal expansion curves of the glaze and body is at a maximum, also because the glaze is about to undergo the glass transition phase.

At the beginning of the cooling phase, the glaze is liquid and follows the body contraction without developing tension. It is possible to identify the coupling temperature (Tc) at 700°C. The glaze and body chemically interact and are bonded together. Tensions are released above Tc, while tensions are built up below Tc during cooling. It could then be argued that during cooling, the curve should be symmetrical with the heating curve. This is wishful thinking, however, because the flexion follows a different path and the system does not come back to planarity. This phenomenon is primarily due to the differences between the CTEs of the glaze and body that are greatly enhanced by the newly formed interface. The glaze has become rigid enough to build up tension, causing flexion of the sample.

In this example, it is easier to identify the coupling temperature during cooling because a rapid variation of inclination in the flexion curve occurs. During the heating phase, the coupling temperature is less evident and more difficult to be identified. It is also possible to see that the curve is below the zero-line (concave specimen) during the whole heating phase and for a part of the cooling phase (from 1,100 down to 550°C). If we assume that for temperatures higher than the Tc the glaze is not able to build up tension, one might ask why the tile is concave at this point and not planar, as may be expected. A simple model has been developed to provide an answer to this question: a planar system composed of a spring (glaze) and a thin plate (body) is associated with the glazed tile bar. Two cases (spring in tension and spring in compression) were analyzed.

Figure 6 shows that, if the spring in compression is uncoupled from the plate, the plate bends downward (concave system) because it was the spring compression keeping the system planar. If the spring in tension is uncoupled from the plate, the plate bends upward (convex system) because it was the spring tension that kept the system planar. From this example, we can extrapolate that if the glaze is initially in a state of compression when the fired glazed tile sample is subjected to temperatures higher than the Tc, the tile will undergo a deformation that leads to a concave shape. Conversely, if the glaze is initially in a state of traction when the fired glazed tile sample is subjected to temperatures higher than the Tc, the tile will undergo a deformation that leads to a convex shape.

Avoiding Defects

A quantitative study of the state of tension established between glaze and body after firing is fundamental in order to prevent problems that occur in glazed products (e.g., delayed crazing or serious planarity defects). To obtain the final result, it is necessary to translate the glaze’s thermal expansion curve so that it coincides with the body’s thermal expansion curve, as well as the corresponding coupling temperature. The two curves, after the translation, do not coincide anymore at the origin (room temperature). This difference is indeed the result of the traction or compression that has been established between glaze and body immediately after firing.

The fleximeter is crucial in order to post-analyze products that are prone to planarity defects after firing. A sample can be directly taken from production and analyzed above its coupling temperature; the bending curve during cooling will reveal how much tension the mechanical system undergoes.