Non-Destructive Characterization of Conservation Materials in Cultural Heritage Restoration Using Terahertz Time–Domain Spectroscopy: A Database Approach

The identification of conservation materials in cultural heritage objects is essential for understanding their history, condition, and appropriate treatment strategies. However, many established analytical techniques require sampling or provide limited access to subsurface information in complex multilayer structures. Terahertz time-domain spectroscopy (THz-TDS) has emerged as a promising non-destructive technique for probing dielectric materials, yet its broader application in heritage diagnostics remains constrained by the limited availability of systematic reference data for conservation materials. In this study, the THz optical properties of a representative range of conservation materials commonly used in restoration practice were systematically investigated using THz-TDS in transmission geometry. The examined materials include collagen-based animal adhesives, natural and synthetic resin varnishes, waxes, solvent-based acrylic resins, poly(vinyl acetate) (PVAc) adhesives, and acrylic and styrene–acrylic dispersions. Samples were prepared under standardized conditions to enable reliable extraction of refractive index and absorption coefficient spectra. The results reveal distinct material-dependent THz responses, allowing classification into low-loss (waxes), intermediate (varnishes and acrylic resins), and high-loss (animal glues and PVAc) materials. While partial overlap is observed between certain material classes, the combined analysis of refractive index and absorption coefficient significantly improves material discrimination. The resulting dataset provides a reference framework for interpreting THz measurements on cultural heritage objects. Our findings demonstrate the potential of THz-TDS as a non-invasive diagnostic tool for identifying conservation materials within multilayer structures. Furthermore, the generated dataset provides a foundation for advanced data-driven approaches, including machine learning–based classification, supporting the development of automated and reliable material identification strategies in heritage science.