(12)
At increasing temperatures, the higher valences of polyvalent ions become increasingly unstable (see Chapter 5.6) so that the fining reaction actually reads
(13)
The release of oxygen bubbles reaches its maximum at about 1300 °C and extends beyond 1400 °C. The negative side effect of this procedure is the formation of the NOx pollutant.
A simple calculation will finally explain why experience and empirical knowledge still play the predominant role in the allotment of fining agents. As used in the batch in Table 4, a mass of 4 kg of Na2SO4 represents 56.3 mol of SO2, which, at 1400 °C, 1 bar, would fill a volume of 7.6 m3. Now, 1 ton of melt, by contrast, fills 0.4 m3 only. Obviously, only a very minor part of the nominal SO2 ends up in gas bubbles otherwise a foam instead of a clear melt would be obtained. The major part of SO2 is in fact lost during batch melting, by evaporation from the melt surface, or is retained in the glass. Thus, the proper allotment of fining rests on the small difference between sulfate input and the above losses. One of the rare attempts to perform a detailed sulfur balance of a glass furnace revealed that approximatively 0.25–0.3 kg of the sulfate added per t of glass are released in the form of fining bubbles [6].
5.3 Homogeneization
After the fining process, the melt is cooled down and homogenized thermally in a steady way. Small residual bubbles resorb themselves because the solubilities of most volatile species strongly decrease with increasing temperatures (Chapter 5.5). For this reason, care has to be taken to prevent local temperature rises from happening during the homogenization process otherwise the so‐called reboil bubbles would form in the melt and could not be removed in any way. Among dissolved gases, N2 distinguishes itself by its decreasing solubility with decreasing temperatures. Thus, N2‐containing bubbles escaping the fining process appear as very tiny bubbles called seeds in the final glass. Their number per unit mass of glass represents an important quality criterion. In container glass, a few tens of seeds per 100 g of glass are accepted. Float glass requires a much higher quality (one visible defect per 20 m2 already is considered a high defect density) and hence, much longer dwell times (approx. 1.5–2 days vs. 1 day for container glass) in the melting compartment.
6 Energetics of Glass Melting
The amount of energy involved in the fusion of glass is an issue of great interest to the glass industry. Referring to comprehensive quantitative treatments ([7, 8] and Chapter 9.8), we will give only a brief sketch of this issue within the scope of this chapter. The approach rests on the fact that, at constant pressure, the heat (enthalpy) transferred to or drawn from a system is thermodynamically the variation of a state function: as such, the intrinsic energy demand depends only on the initial and final states of the system and it can be determined without any consideration of what is going on along the process road.
The initial enthalpy state is given by the sum of standard enthalpies H°i at 25 °C, 1 bar, of the individual raw materials i, weighted by their respective amounts mi in the batch:
(14)
The final enthalpy is given by the standard enthalpies of the batch gases g, H°GASES = ∑ mg·H°g, and of the glass, H°GLASS, plus the heat content ΔH(Tex) of the glass at the exit temperature Tex. The standard enthalpy difference between inputs and products constitutes the chemical energy demand
(15)
The heat content of the melt at Tex is given by ΔH(Tex). For convenience, all enthalpy values are inserted in absolute figures, disregarding the minus sign given in thermochemical tables. The overall intrinsic heat demand Hex (exploited heat of the process) is given by
(16)
where yCULLET denotes the weight fraction of cullet per amount of glass produced.
It is true, real raw materials typically do not contain their main mineral phase only, but also contain minor amounts of side minerals. For example, a real quartz sand may contain, beside its main phase quartz, minor amounts of feldspar minerals, magnetite, spinel, etc.; a natural dolomite is typically composed of different minerals forming solid solutions in the system Ca–Mg–FeII–CO3 with an overall composition not too far from the pure phase CaMg(CO3)2. An accurate determination of the enthalpy values H°i of real raw materials would thus require the evaluation of multicomponent phase diagrams. However, such an approach would hardly be accepted by the technological community. Beyond this, the gain of accuracy against a simpler approach is minor only. Thus, with the reservation to a more rigorous treatment [7, 8], only the enthalpy values H°i of pure raw materials are given here in units of MJ/kg:
| Raw material i | Enthalpy H°i in MJ/kg |
| Pure quartz sand | 15.150 |
| Pure albite (NaAlSi3O8) | 14.952 |
| Pure dolomite CaMg(CO3)2 | 12.549 |
| Pure calcite CaCO3 | 12.058 |
| Soda ash | 10.659 |
| Sodium sulfate | 9.782 |
| Carbon | 0.000 |
| Calumite® | 13.561 |
For the batch gases, the following values hold:
CO2: 8.941; H2O: 13.422; SO2: 4.633; O2: 0.000.
The energy calculation for the real glass composition of Table 2 (where the tiny amount of TiO2 has been allotted to SiO2) is summarized in Table 5. The position of the glass composition in the phase diagram in units of kg of equilibrium compounds per t of glass is found by the following simplified procedure:
NAS6 = 51.440 Al2O3 – 55.697 K2O,
KAS6 = 59.102 K2O,
hm = 6 Fe2O3,
FS = 7.345 Fe2O3,
MS = 24.907 MgO,
NC3S6 = 35.112 CaO,
NS2