Proper sampling of raw materials [4] thus is needed to guarantee their conformity with regard to possible geological heterogeneity and product variability at the quarry level. It relies on suitable quartering techniques to obtain a true fingerprint of the mineralogy of all raw materials from which incorporation of harmful species may be ruled out or at least minimized. In other words, heavy‐mineral content is an overwhelming and crucial specification concerning the physical and chemical properties that must be guaranteed by a producer of raw materials, especially when fabrication of a new glass has to be tested. Not complying with these specifications can generate long‐lasting yield drops and large financial losses for the glassmaker. Since glassmakers constantly need to diversify and secure their supplies of raw materials, heavy‐mineral characterization must routinely be operated by well‐equipped internal or academic laboratories.
Of course, the bulk chemistry must also be determined on a daily basis at the plant by XRF, ICP‐MS, or wet chemistry (cf. Chapter 5.1) to monitor the variability of moisture and oxide content, especially for multielement raw materials, and thus to allow batch adjustments needed to keep the glass recipe constant to be calculated (cf. Chapter 1.3). As indicated above, the PSD, LOI, and COD parameters must in addition be included in the almost daily control of raw materials at the plant. In this way it is possible to anticipate possible drifts away from the targeted specifications of the raw‐material feed. As for the overall meltability, energy demand, and expected quality, they may be tested less routinely through differential scanning calorimetry (DSC) measurements, while it should be compulsory to test the actual batch incrementally, first in the laboratory (few kg), then in a pilot furnace (~1 ton), and finally at the industrial scale (<1000 tons).
3.3 Impurity‐related and Other Melting Defects
Melting quality first and foremost depends on appropriate digestion rates. Batch stones can, for instance, readily occur when market requirements push glassmakers to increase pull rates and, consequently, to reduce the average residence time of the raw materials in the furnace. They also form in case of errors in batch calculation or of scale malfunction. The problem is illustrated with the overall texture and microstructure of the silica batch stone shown in Figure 3. Noteworthy are former quartz grains, whose shape have been preserved although they have been totally replaced by cristobalite (as a pseudomorph) from which newly formed tridymite lath‐shaped crystals have grown radially in a groundmass of silica glass [10]. These textural features are typical of an insufficiently dispersed quartz sand within the batch. Regardless of the actual pull rate, the overall moisture content and distribution can provoke the formation of lumps when grains stick together at the batch plant and quickly sinter at the dog‐house level, preventing natural convection within the melter from properly stirring the melt under formation. Moisture is of special concern in regions facing very wet or very cold (ice) seasons. Preheating or protecting the raw materials yard can then be an effective, but somewhat costly solution.
Figure 3 Silica batch stone in a soda‐lime silica glass, resulting from incomplete digestion of a lump of quartz grains as viewed under an optical microscope (a) and as observed in a thin section under transmitted light (b), where rounded grains of cristobalite formed as quartz pseudomorph are sluggishly digested while generating radially growing tridymite laths in a vitreous groundmass showing an overall open‐porosity.
Quartz batch stones can also result from an inadequate overall PSD when other raw materials contain up to several weight % of free silica as impurities, whose size distribution differs from that of quartz sand. Limestone, dolomite, and feldspars are typical examples, the two carbonates having a dmax for quartz as high as 2 mm or more (Figure 2). Given the aforementioned digestion rates of quartz, such large grains will end up as unmolten stones in the production process.
Figure 4 Undissolved chromite crystal in a soda‐lime silica glass as seen under a binocular microscope (a) and in a polished thin section photo under reflected light (b). The chromite grain shows a preserved FeCr2O4 core, exsolution lamellae, and radially growing eskolaite Cr2O3 laths in the outer shell.
Refractory minerals, of course, raise special difficulties as illustrated in Figure 4 by an incompletely dissolved chromite inclusion. Its core is preserved as FeCr2O4, but it is surrounded by newly formed laths of eskolaite [Cr2O3] radially growing from it according to the decomposition reaction:
(1)
Although Fe2+ then diffuses in the melt, the highly refractory eskolaite crystals (melting at 2435 °C) passivate the chromite core, which will thus remain throughout the production process.
Incomplete digestion can alternatively cause the existence of vitreous inclusions called knots in the glassmaker jargon. The example shown in Figure 5 is that of an mm‐sized pocket of alumina‐rich glass within the normal soda‐lime silica matrix, which represents the ghost of a feldspar crystal. As already stated, feldspar minerals melt at temperature below 1200 °C, therefore early in the process, i.e. already at the dog‐house level. When they do so, they generate an alumina‐rich liquid phase whose viscosity of 105 dPa s at 1000 °C is 10 times higher than that of the soda‐lime silica glass [11]. At high pull rates this difference prevents rapid enough interdiffusion from taking place to ensure complete mixing between the two liquids, whence the presence of knots. A simple solution to avoid them when Al‐carriers are feldspars thus is to control the PSD of these minerals. But, with very different PSD ranges, feldspars can also be present as impurities in a variety of materials such as quartz-sand, limestone, and dolomite. Likewise, a proper PSD helps minimizing the presence of alumina‐rich knots in the final product when the Al‐rich phonolite and nepheline syenite rocks are used as raw materials.
Figure 5 A feldspar knot with about 20 wt % Al2O3, enclosing bubbles (gas inclusions) in a soda‐lime silica glass as seen under a binocular microscope.
Even cullet may be a source of defects,