Although there is a common consensus that aluminium in silicate melts is generally stabilised in tetrahedral coordination by association with other metal cations, its structural role in melts in which there is a stoichiometric excess of Al over other cations has been the subject of some debate. It has been suggested that such excess aluminium may be octahedrally coordinated in a network modifying role, or alternatively associated with a three coordinated oxygen in a configuration known as a tricluster. Although NMR spectroscopy and MD simulations have shown the presence of high coordinate aluminium in very peraluminous melts and glasses, the situation for compositions close to the metaluminous, charge-balanced join remains controversial.
As aluminium is substituted for metal cations at fixed silica content a change in the structural role of Al may be expected to occur when Mn+/nAl = 1 (i.e. the metaluminous join, where M = Na, K, Mg, Ca, etc.). This change should be manifest in variations of melt viscosity, a property which is extremely sensitive to melt composition and structure. Shear viscosities of melts along various silica isopleths with mol (Mn+/(Mn++nAl)) in the range 0.60 to 0.40 have been determined using the concentric cylinder method, with M = Na, Ca and Mg. The experimental set-up was designed to maximize the precision of the measurements, permitting the accurate constraint of the position of a viscosity maximum.
At fixed silica content in the system Na2O-Al2O3-SiO2
(NAS) it is found that a viscosity maximum occurs consistently within the
peraluminous field. In the range 45 to 82 mol% SiO2 the greatest
displacement of the viscosity maximum from the metaluminous join is at
67 mol%. Firstly, this result implies that not all aluminium is charge
balanced by sodium in tectosilicate melts. Secondly, a maximum in displacement
at 67 mol% SiO2 is difficult to reconcile with the presence
of octahedral aluminium, but may be explained by the presence of triclusters
consisting of one aluminate and two silicate tetrahedra, whose formation
may be described by the homogeneous equilibrium:
Preliminary experiments in the system MgO-Al2O3-SiO2 do not show a viscosity peak in the peraluminous field, but suggest that a viscosity maximum occurs within the field Mg/2Al>1. This result is inconsistent with the presence of triclusters, but may suggest the presence of octahedral aluminium in the case of magnesium aluminosilicates, as is indeed suggested by NMR spectroscopy of some very rapidly quenched magnesium aluminosilicates. Viscosity measurements along the 67 mol% SiO2 isopleth in the system CaO-Al2O3-SiO2 show two maxima, one within the peralkaline field, and the other within the peraluminous field, that within the peralkaline field disappearing with decreasing temperature. The peak within the peraluminous field shows a greater displacement from the metaluminous join than the Na2O bearing system, implying that triclusters are more abundant in the presence of Ca than Na. The presence of such triclusters may be of importance, not only for the physical properties, but also for the dynamics of viscous and diffusive flow. For example, formation and destruction of triclusters may play the role in highly polymerized aluminosilicate melts that the formation of 5-coordinated Si from non-bridging oxygens is known to play in depolymerized melts.