Determination of a discrete relaxation spectrum from dynamic experimental data using the Padé-Laplace method

The Pade-Laplace method using the Laplace transform and the Pade approximants has been used to recover a discrete relaxation spectrum from dynamic oscillatory data in the molten state for polymers. The method detects by its own stability the number of modes of the spectrum and has been shown to be able to provide spectra with a low number of modes. These spectra are found to be sufficient to obtain an accurate description of the linear viscoelastic behavior in a large frequency range from the terminal zone to the plateau zone. Some intrinsic parameters of the numerical technique have been extensively studied in order to refine it. Its applicability is shown in the case of polystyrene having monomolecular, bimodal or broad molecular weight distributions. It can easily be implemented on a personal computer and the computing time is of the order of a few seconds.