Viscoelastic relaxation of supercooled liquids. II

Abstract In addition to the six liquids for which the results of viscoelastic measurements are reported in part I, studies have also been made on a further eight liquids which differ widely in molecular composition, namely: squalane, tri(β-chloroethyl) phosphate, tri(o-tolyl) phosphate, tri(m -tolyl) phosphate, tris(2-ethyl hexyl) phosphate, tetra(2-ethyl hexyl) silicate, m-bis(m-phenoxy phenoxy) benzene and bis(m-(m-phenoxy phenoxy) phenyl) ether. It has been confirmed for each liquid that the limiting shear modulus, G∞, varies with temperature according to the relation previously found, 1/G∞ = 1/G0 + C(T - T0). Plotting the quantities RL(ρG∞)1/2 and XL/(ρG∞)1/2 against log10(ωƞ/G∞), where RL and XL are the resistive and reactive components of the shear mechanical impedance, shows that the results for six of these liquids are indistinguishable within experimental error. Moreover, the curves drawn through the experimental points represented in this manner are identical with the corresponding curves found for five of the liquids for which results are described in part I. The results for these eleven liquids show a striking agreement. A new theoretical model is put forward from which the viscoelastic properties of pure liquids in the supercooled region can be predicted within experimental error. This is regarded as a limiting case for liquids which follow the free volume description of steady flow viscosity. In the light of available evidence, it is envisaged that the viscoelastic relaxational behaviour approaches that represented by a single (Maxwell) relaxation process as the degree of co-operative motion of the molecules decreases. Only three of the fourteen liquids investigated show a behaviour which does not follow the theoretical prediction. The observed deviations are not great but are outside the limit of experimental error. Measurements on mixtures of liquids which individually exhibit behaviour in agreement with theory show that an impurity content of less than 2% can result in similar departures from theory. The lack of agreement for the three liquids in question is therefore attributed to impurities in the samples measured.