Component-Resolved Diffusion in Multicomponent Mixtures. A Case Study of High-Field PGSE−NMR Self-Diffusion Measurements in Asphaltene/Naphthenic Acid/Solvent Systems

The diffusion of a multitude of different-sized molecules in various asphaltene mixtures has been measured by means of the pulsed gradient spin−echo nuclear magnetic resonance (PGSE−NMR) self-diffusion technique. The multicomponent mixtures studied contained mixtures of polydisperse asphaltenes and monodisperse naphthenic acid in solvent. Spectra obtained from a high-field PGSE−NMR self-diffusion experiment on these systems showed that the peaks from all components were almost completely overlapping, making a “standard” least-squares fit approach for evaluating individual self-diffusion coefficients from the decrease in signal intensity quite difficult. However, to study the details of the molecular interactions that occur between asphaltenes and naphthenic acids, the experimental data must be such that component-resolved information is obtained. Hence, with the purpose to obtain a robust evaluation procedure, with regard to obtaining component-resolved diffusion coefficients, a “standard” least-squares fit approach was tested against a global least-squares approach. The global least-squares evaluation was performed using a program called CORE (component-resolved spectroscopy). CORE not only fits predetermined functions to specific peak intensities or areas obtained from the diffusion experiments, but also considers the full information in the spectra. That is, the intensity at each frequency in the spectra is considered and the band shapes for all components in the mixture can be calculated. The calculated band shapes are thus a good test of the validity of the results for the diffusion coefficients. In this work, the band shapes were calculated and the diffusion for each individual component in the samples was extracted. The band shapes were observed to be in close agreement with the spectra obtained for the pure components. In addition, the diffusion coefficients, as obtained from CORE and from the least-squares fit method, were in close agreement, although some discrepancies were noted.