A recent study of post-campaign AZS refractory blocks from three types of glass melting furnaces is providing a better understanding of the link between AZS refractory corrosion and glass defects.

Glass defects such as bubbles, stones, knots and cords originate from many sources, including poor melting and/or refining of glass, as well as refractory quality and degradation. The published literature often cites alumina-zirconia-silica (AZS) refractory exudation as the leading cause of knot and cord defects. However, one study of the origin of knot defects in TV glass showed that AZS refractory exudation is only a short-term source of glass defects, while AZS refractory corrosion, especially in the superstructure, is a more potent and long-term source of defects.¹

Very little information exists in published literature on defect chemistry and frequency as a function of furnace age and AZS reuse, so researchers at Vesuvius Monofrax, Inc. attempted to correlate glass defect chemistry with AZS refractories of varying ages. They analyzed post-campaign AZS refractory blocks from three types of glass melting furnaces: a soda-lime container furnace, a soda-lime tubing furnace and a lead silicate TV funnel furnace. The results of this study are providing a better understanding of the link between AZS refractory corrosion and glass defects.

Soda-Lime Container Furnace, 10-Year Campaign

Figure 1 shows an AZS superstructure (left and inset) and a glass contact refractory block (right) taken from a soda-lime container furnace following a 10-year campaign. While the glass contact block reveals rounded edges from corrosion and a shiny surface due to glass adhering to the refractory, the superstructure refractory appears to be dry on the surface, has relatively sharp edges and shows a whitish crust on the entire exposed hot face. The glass contact block appears to have come from below the metal line.

The holes in the blocks are from drilling core samples for characterization. Polished sections for microscopy were prepared from the core samples. The chemistry and microstructure were analyzed as a function of depth using SEM/EDS techniques. In addition, physical properties such as the bulk density and the apparent porosity were also measured as a function of depth.

The lower portion of Figure 2 shows SEM/BSE photomicrographs of the glass contact interface and superstructure hot face. The microstructure of both types of samples shows a near absence of the crystalline alumina phase. Both the superstructure and glass contact AZS samples were found to contain a nephelitic zone where the zirconia is no longer in solution in the glassy phase; instead, it exists as discrete zirconia crystals. This nephelitic zone extends to greater depth in the superstructure than in the glass contact interface.

The chart on the upper portion of Figure 2 shows the matrix phase chemistry as a function of depth for both the glass contact and superstructure AZS refractory samples. Both samples exhibit similar changes in chemistry-i.e., an increase in the concentration of the alkali/alkaline earth species, an increase in the alumina concentration and a decrease in the concentration of silica. However, the superstructure refractory sample shows a greater depth of chemical change than the glass contact sample. An example of this is the depth of the nephelitic zone, mentioned above, which was found to be greater in the superstructure sample than in the glass contact sample.

The results shown in Figure 2 suggest that the glass contact refractory corrosion lessens with time, while the superstructure corrosion continues over the life of the furnace campaign.

Soda-Lime Tubing Furnace, Five-Year Campaign

Figure 3 shows a photo of an AZS glass-contact sidewall block obtained from a soda-lime glass tubing furnace following a five-year campaign. The inset photo shows a viscous knot glass defect also obtained from this tubing furnace. In addition to analyzing the chemical, physical and microstructural changes in the refractory block, researchers also analyzed the glass defect to determine if its source was the glass-contact AZS refractory.

The SEM/BSE photomicrograph shown in the lower portion of Figure 4 displays the microstructure of the AZS block at the glass/refractory interface. The changes in the AZS microstructure are similar to those found in the soda-lime container furnace. The crystalline alumina phase is essentially absent in the micrograph.

The chart in the upper portion shows the chemical analysis of the matrix phase (glassy in the as-is AZS refractory) as a function of depth in the AZS block. The chemistry of this tubing glass is similar to that of the soda-lime container glass described previously, and the chemistry profile of the AZS block was found to be similar to that seen in the container glass tank. The alkali/alkaline earth and alumina contents increased, while the silica content decreased in the matrix toward the glass/refractory interface. This resulted in the formation of a nephelitic matrix phase.

The chart also shows the chemistry of the viscous knot defect, represented by open symbols at an arbitrary position of -5 mm from the glass/refractory interface. The lack of ZrO₂ in the cord chemistry strongly suggests a non-AZS cord source.

Figure 5 compares the depth of corrosion in the glass contact AZS blocks taken from the container and tubing furnaces. Though the corrosion depth is greater in the container furnace, this could be due to the differences in the campaign durations (10 years for the container vs. 5 years for the tubing), location in the furnace and temperature.

TV Funnel Furnace, Four-Year Campaign

The third post-campaign AZS refractory sample was obtained from a TV funnel furnace after approximately four years of service. A core sample, B1 as shown in Figure 6, was drilled from the center of the charge end superstructure wall. The extensive rundown seen on the wall is most likely due to the corrosion of the AZS blocks, which was accelerated by the inevitable presence of batch dust in the charge end area. A cross section of the core sample (top right photo) shows a tear running from the hot face into the block interior, and a change in the color of the refractory.

The chart in Figure 7 displays the concentration of all alkali and alkaline earth species in the B1 core sample as a function of depth. The chemical analysis was performed using SEM/EDS in the line scan mode at a low magnification. This method allows bulk chemical analysis at a known depth into the sample. The chart compares the alkali and alkaline earth concentration of the exposed AZS sample to the concentration of the same species in the as-is AZS refractory. There is a significantly higher concentration of the analyzed species up to at least 50 mm.

A series of reflected light photomicrographs were collected from the polished samples; two of these photomicrographs are shown in Figure 8. A complex array of phases has formed due to the chemical changes in the AZS block. The in-diffusion of alkaline and alkaline earth species (potassium, sodium and magnesium) with the concomitant dissolution of crystalline alumina into the matrix has resulted in the creation of new crystalline phases such as nepheline, kalsilite, leucite and beta-alumina.

Summary of Results

Table 1 summarizes the depth of corrosion measured in all the AZS samples discussed in this article. While the numbers shown in the table are not meant to be used for calculating the rate of corrosion of glass contact and superstructure AZS refractories, the trend observed here supports the conclusion that superstructure AZS undergoes a greater degree of chemical change than the glass contact AZS below the metal line. All other variables being equal, this data therefore suggests that superstructure corrosion can continue as the furnace ages, while glass melt contact corrosion (below the metal line) can slow down with the age of the furnace.

In addition to studying the corrosion behavior, researchers also measured changes in the physical properties of the post-campaign AZS samples. The level of apparent porosity was, in general, higher than that seen in the as-is AZS. This increase is most likely due to the formation of new crystalline phases (that can be higher in density and thus lower in volume) at the expense of the glassy phase, and also from liquid phase rundown from the refractory surface into the glass bath.

AZS Corrosion and Glass Defect Formation

Figure 9 explores a correlation between the AZS refractory corrosion and knot/cord-type glass defects. To create this chart, researchers plotted the chemistry of all knot and cord defects analyzed at the Monofrax Technical Center within the last five years. The chart compares the ZrO₂ concentration with the Al₂O₃/ZrO₂ molar ratio of the defects. When assigning the most likely origin of the defect-i.e., superstructure or melt contact AZS corrosion-the researchers used the chemistry of the AZS refractory samples discussed in this article, as well as many other samples available in their database.

The defect represented by a triangle symbol in Area 1 is believed to have definitely originated from the melt contact corrosion of AZS refractories. The defects represented by square symbols in Area 3 are believed to have definitely originated from the superstructure corrosion of AZS refractories. Defects noted in Areas 2 and 4 represent some uncertainty about the source; however, the defects in Area 2 are most likely from an AZS melt contact source, and the defects in Area 4 are most likely from an AZS superstructure source.

Since the defects shown in this chart came from furnaces melting many types of glass chemistries, and the majority of the defects appear to have originated from superstructure AZS corrosion, it seems logical to conclude that glass defect formation continues as a given glass melting furnace ages.

Minimizing Defects

Based on the studies described in this article, it is evident that both short- and long-term corrosion mechanisms are similar. Glass melt contact refractory corrosion can lessen with time due to boundary layer formation and the effect of external cooling at the metal line. Superstructure refractory corrosion, however, can continue through the entire campaign duration, as evidenced by the soda-lime container tank study, where the corrosion depth was ~50 mm glass contact and ~150 mm superstructure.

Post-campaign refractory evaluation has also shown an increase in apparent porosity, which may be due to the formation of new phases and/or a matrix (liquid) phase rundown. The chemistry of both knots and cords appears similar to the AZS hot face chemistry following corrosion. Though both glass contact and superstructure corrosion products can lead to knot/cord defects, superstructure corrosion is a more potent and long-term source of defects. Additionally, used AZS can contain new crystalline phases such as nepheline, kalsilite, leucite, beta-alumina and zircon. And a mismatch in the coefficient of thermal expansion (CTE) with unaltered AZS may lead to spalling, which can also cause defects.

As mentioned earlier, the glassy matrix phase in AZS (which is effectively 1/3 of the total volume) provides large pathways for the corrosive alkaline and alkaline earth species to diffuse into the body of the refractory. This in-diffusion promotes the dissolution of crystalline alumina, resulting in an expansion of the glassy phase volume. The data from post-campaign AZS superstructure blocks show significant chemical alteration of the glassy phase in up to several inches of the block thickness.

Given that the majority of the knot and cord defects analyzed at Monofrax are similar in chemistry to that of the AZS glassy phase following superstructure corrosion, and that AZS superstructure corrosion can progress over the life of the furnace campaign, it is reasonable to conclude that superstructure AZS corrosion is an ongoing source of glass defects. However, the rate of defect generation is a more complex issue and depends on many other factors besides refractory degradation. These factors include furnace temperature profile, throughput and furnace exhaust control, which were not analyzed in this study.

Obtaining a correlation between glass defect frequency and furnace variables would require meticulous recordkeeping of the furnace process conditions and defect levels throughout the furnace campaign. Glass manufacturers can either engage in this type of long-term study, or find alternatives to reduce defects based on the current understanding of refractory degradation.

There is no doubt that all AZS refractories experience an expansion of the glassy phase volume due to superstructure corrosion and can therefore serve as a source of liquid phase rundown, promoting glass defects. Therefore, the best solution would be to avoid using AZS refractories in the superstructure lining of airfuel furnaces altogether.

One suitable alternative is a fusion-cast alpha-beta alumina refractory, which contains a very small amount of crystalline boundary phase (~2% by volume) bearing nepheline-type chemistry. Comparative studies of AZS and alpha-beta alumina refractories superstructure corrosion in airfuel and oxyfuel furnaces have shown significantly lower chemical alteration of the alpha-beta alumina than that seen in AZS. Furthermore, over the last 10 years, alpha-beta alumina refractories have been successfully used in the crown and superstructure of oxyfuel-fired glass melting furnaces, showing excellent physical and chemical stability over multiple campaigns.

By understanding how glass defects occur, manufacturers can take the appropriate steps to minimize these defects and improve glass quality.

For more information about glass furnace refractories, contact Vesuvius Monofrax, Inc., 1870 New York Ave., Falconer, NY 14733-1797; (716) 483-7200; fax (716) 661-9296; e-mail amul_gupta@us.vesuvius.com; or visit http://www.monofrax.com.

Reference

1. Proceedings of the 62nd Conference on Glass Problems, October 2001, pp. 59-82.