Reaction (1) can also be considered as a description of chemical equilibrium among oxygen ion species, and has long played a prominent role in models of the thermodynamics of low‐silica metallurgical slags. Conventional views have suggested that “unreacted” or “free” oxide ions are probably only abundant in low‐silica glasses and/or in systems with highly electronegative modifiers such as Sn2+ and Pb2+. Oxygen ion speciation in some silicate glass compositions can be measured with methods such as XPS and 17O NMR, or estimated indirectly through techniques that provide information on silicate anionic speciation such as Raman spectroscopy and 29Si NMR (see Section 7.2). Some such studies have suggested the presence of a few percent of “free” oxide ions in compositions well outside the range where they are required by stoichiometry alone, e.g. at silica contents >33.3 mol% in the MgO–SiO2 binary. As containerless melting and quenching methods have made the formation of very low silica glasses possible (even into the “sub‐orthosilicate” range, e.g. lower silica than in Mg2SiO4), it has become possible to quantify more clearly this most basic aspect of silicate structure [6].
In “modified,” ambient‐pressure silicate glasses, only tiny fractions of SiO5 groups have been detected in a few alkali silicates. However, high pressures, or high P2O5 contents, can lead to the formation of substantial fractions of both SiO5 and SiO6, which will contribute to the overall network disorder as well as to density increase. For example, both an unusual high‐pressure crystalline phase of CaSi2O5 and its glassy equivalent clearly have all three Si coordinations (Figure 5). A number of in‐situ, high‐pressure studies, particularly by Raman spectroscopy, have suggested that considerable structural relaxation, and reversion to lower network cation coordination, can take place on decompression of a glass, even at ambient temperature. A few of these have probably also taken samples above Tg at high pressure. The small cationic radius and high charge of P5+ makes P2O5 another well‐known network‐forming oxide. Over wide ranges of composition, two‐component and more complex phosphate liquids are stable and can be quenched to glasses. These have been extensively studied, particularly by vibrational spectroscopy and 31P MAS NMR. All phosphorus is present in PO4 tetrahedra. As in silicate glasses, these are linked together to form chains of varying length as well as more complex structures. The roles of BO and modifier oxides in phosphate glasses are analogous to those in silicates, as are the considerations of anionic speciation discussed in Section 7.2.
Figure 5 Silicate structural groups in a high‐pressure, triclinic crystalline phase of CaSi2O5 and in its isochemical glass quenched from the melt at 10 GPa, showing the correspondence of signals for Si with 4, 5, and 6 oxygen neighbors in 29Si MAS NMR spectra.
Source: Modified from [7].
4 Modifying the Network: Borates and Germanates
Contrary to what is found in ambient‐pressure silicates and phosphates, a different type of network modification takes place when oxides of low‐valence cations are initially added to B2O3 or to GeO2, facilitated by the energetically “easy” transitions of network cations between two (BO3, BO4) or even three (GeO4, GeO5, GeO6) coordination states. Instead of only forming NBOs, the added oxide ion serves primarily to increase the coordination number of the network cation so that the network remains fully connected by BO, if the definition of the latter is expanded to include linkages with the higher‐coordinated network cations. (It is important to note, of course, that oxygen bridges between network cations may be energetically quite distinct, and have differing implications for bulk properties, as the network cation coordination varies.) If NBOs do form, their concentrations are much lower than produced in the corresponding silicate equilibria. The “modifier” cations are coordinated primarily by BO, some of which will have partial negative formal charges, e.g. −1/4 on the BO linking a BO3 with a BO4 group. This mechanism (Figure 4) predominates up to about 20–30 mol% modifier oxide, at which point the formation of “normal” NBOs, as in silicates, begins to become important. At least part of this turnover may result from the difficulty of packing enough low‐charge modifier cations around BOs with higher formal charges, i.e. −1/2 on the link between two BO4 groups, and in turn this can be affected by the cation field strength and dilution of the borate network by silica. At higher modifier contents, much or even most of the network cation coordination returns to the lower state, BO3 or GeO4. These compositionally induced transitions in the network cation coordination are generally mirrored in the structures of the binary crystals, and result in strongly nonlinear property–composition relationships in both the melts and glasses, e.g. density and glass transition temperature.
This mechanism is most precisely defined in alkali borate glasses, for which 11B NMR has long been applied to measure directly BO3 and BO4 contents [8]. Raman spectroscopy can also detect this coordination shift, and can be more readily applied at high temperatures and pressures. The structural transition with composition can be symbolized by the reaction, which incorporates the reaction of O2− with BOs to form NBOs:
(2)
This reaction can also be taken as a statement of chemical equilibrium among melt species. Shifts with temperature have been determined from both in‐situ, high‐temperature vibrational and NMR spectroscopy and studies of glasses prepared at different cooling rates and thus with different fictive temperatures [9]. The lower coordination state (left hand side) is generally favored at higher temperature, meaning that the enthalpy change for the reaction as written is negative. In boron‐rich systems, this coordination change can be a major contributor to the overall configurational heat capacity and enthalpy of the liquid; changes in the abundances and mixing of the boron coordinations will clearly affect the configurational entropy as well. At least in borosilicate glasses, modifier cations with higher field strengths tend to favor the formation of NBOs and thus lower boron coordination numbers. Spectroscopy on quenched, decompressed glasses has shown that this mechanism leads to boron coordination increase at high pressure [10]; a few in‐situ studies by X‐ray and other methods have observed this process more directly.
Analogous structural transitions that take place as modifier oxides are added to GeO2. Alkali germanate glasses and melts have density maxima at roughly 15–20% M2O, the compositions near which crystal structures are made up of mixtures of GeO4, GeO6,