Molecular Characterization of Aged Bitumen with Selective and Nonselective Ionization Methods by Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. 2. Statistical Approach on Multiple-Origin Samples

Abstract

Bitumen is a very viscous and complex mixture primarily used as a binder for asphalt construction. During its service life, this binder is subjected to oxidative aging, which leads to asphalt stiffness and thermal cracking of the pavement structure. The behavior of bitumen toward oxidative aging is greatly dependent on its intrinsic properties and particularly on its content of nitrogen, oxygen, and sulfur functional groups. This is of fundamental importance as the origin of the crude will influence its resistance to oxidation after several years of use in a road. It is then essential to correlate the native composition of bitumen and their corresponding oxidized products to explain changes in the physicochemical behavior of asphalt binders. Pressure aging vessel is probably the most widely used technique to simulate long-time aging of bitumen. This artificial aging technique leads to accelerated aging under high-pressure and -temperature conditions. To understand such mechanisms, characterization at the molecular level is required. Nowadays, Fourier transform ion cyclotron resonance (FTICR) is the most-used technique for the analysis of petroleum products. Bitumen samples from three different origins were subjected to PAV aging and analyzed by FTICR-MS. Coupling of electrospray ionization (ESI) and atmospheric pressure photo ionization (APPI) sources were used to provide an exhaustive characterization of the samples. Principal component analysis (PCA), which is a multivariate statistical approach, was used to separate samples according to aging and origin. Indeed, ESI coupled to PCA allowed a significant separation of samples according to their PAV-aging on the PC1 axis. PC1 scores were then used as an intensity scale to carry out a modified double bond equivalent versus carbon number (DBE vs C#) mapping. This technique allowed the characterization of specific aging markers in Oz and OzSy families for each bitumen sample. In addition, PC1 coordinates were used to develop a predictive model by principal component regression (PCR). Significant differences in sensitivity to oxidation were obtained for the three bitumen samples. Concerning APPI results, less significant differences were obtained for aging compared to those obtained by ESI. PCA coupled to APPI was also used for modified DBE vs C# maps, separating samples according to their origin. Significant differences were obtained in each bitumen sample in the different compound classes (HC, S1, N1, O1, etc.). Overall, this study demonstrates the great interest of coupling FTICR MS data to a PCA statistical approach to characterize oxidative aging of bitumen samples according to their origin.