The structures of a wide range of fast ion conducting glasses have bee
n modelled using the reverse Monte Carlo (RMC) method, based on neutro
n diffraction, X-ray diffraction and EXAFS data. The glass structures
typically consist of a 'network' component, e.g., phosphate chains, an
d a 'salt' component. These two components are strongly interlinked. T
here is no evidence that the salt component forms, to any significant
extent; nanocrystalline clusters similar to the equivalent fast ion co
nducting crystals. The main effect of salt doping is to expand the net
work, leading to an increase in the accessible free volume and hence t
he number of conduction pathways. There is a strong empirical correlat
ion between network expansion and increase in conductivity. However sa
lt doping also increases the number of charge carriers and changes the
average cation environment, thus decreasing the activation energy and
increasing the cation mobility, and these effects cannot be separated
from the network expansion. Increased salt doping reduces the number
and strength of network cross links and weakens the glass structure, e
ventually leading to crystallisation.