Terahertz nano-identification of molecular resonances through analytical near-field modeling

Low-energy molecular motions provide chemically specific terahertz (THz) fingerprints, yet diffraction masks their spatial variation within microscopic material domains. Near-field spectroscopy overcomes this spatial limit, but material identification remains challenging because tip-sample coupling transforms the intrinsic dielectric response. Here we combine local complex normalization with constrained analytical prolate-spheroid modeling to recover the cinnabar dielectric response jointly from amplitude and phase. A single Lorentz oscillator accounts for the full complex spectrum across the dielectric, surface, and near-field response chain. This analysis distinguishes the intrinsic 1.12 THz loss maximum from the near-field phase maximum measured near 1.17 THz. Boundary spectra exhibit a red-shifted near-field resonance despite a fixed molecular loss peak, revealing how local dielectric coupling reshapes the measured fingerprint. The framework advances THz nano-identification beyond empirical peak assignment towards physically interpretable molecular and compositional mapping.