Use of multivariate clustering analysis to investigate the physicochemical interactions in bitumen mastics using micromechanical modeling and FTIR spectroscopy

The debate continues regarding whether different mineral fillers and asphalt technologies mitigate or accelerate aging formation in bituminous mastics. Hence, this study was centered on investigating the concurrent impact of aging stages, mineral composition of filler components (e.g., limestone, basalt, and diabase), and asphalt binder technology on the formation and evolution of physicochemical interactions in bituminous mastics. Various bituminous mastics at different aging stages were analyzed using rheological oscillatory testing and Attenuated Total Reflectance Fourier Transform Infrared spectroscopy (ATR-FTIR) measurements to study their chemical characteristics. A generalized self-consistent scheme (GSCS) and Hashin micromechanical models were utilized to estimate immobilized binder volume and adsorbed binder thickness values from rheological data. Additionally, aging-related indices such as carbonyl and sulfoxide indices were calculated from FTIR-ATR results. In addition to the typical comparison of these results, a multivariate clustering approach was employed, comprising Partial Least-Square Regression (PLSR) as a dimensionality reduction protocol and Hierarchical Clustering Analysis (HCA) as a clustering technique. This approach segregated the mastics into four distinct clusters without prior knowledge of their labels. The clustering results grouped variants with similar physicochemical attributes in specific clusters, revealing the mitigating role of various filler types based on the dominant components of each cluster. Based on the physicochemical results obtained from the clustering analysis, limestone filler is expected to play a more significant mitigating role in foamed bitumen mastics, helping to alleviate the effects induced by aging protocols.