Analysis by D L Sidebottom of the dispersive frequency response of the real-part of the conductivity,
σ'(ω), for many alkali phosphate and metaphosphate glasses, using a fitting model
involving a 'universal dynamic response' power law with an exponent
n and a constant-loss
term, led to anomalous n
behaviour that he explained as arising from variable constriction of the local cation conduction
space. In order to obtain adequate fits, he eliminated from the data all low-frequency decreases of
σ'(ω)
below the dc plateau, ones actually associated with electrode effects. Such a cut-off does
not, however, eliminate electrode effects possibly present in the high-frequency part of the
data range. The results of the present detailed analysis and fitting of both synthetic data
and several of his experimental data sets show unequivocally that his anomalous
n
behaviour arose from neglecting electrode effects. Their inclusion, with or without data
cut-off in the fitting model, leads to the expected high-frequency slope value of
n = 2/3
associated with bulk conduction, as required by recently published topological
effective-dimension considerations for dielectric relaxation in conductive systems.
Further, the effects of the inclusion in a full fitting model of series and possibly
parallel complex constant-phase-element contributions, representing electrode and
nearly constant loss effects, respectively, have been investigated in detail. Such
composite models usually lead to best fitting of either the full or cut-off complex data
when they include the semi-universal, topologically based K1 bulk model, one
indirectly derived from the assumption of stretched-exponential temporal behaviour.