A systematic basis set and method benchmark for infrared intensities and dipole moment derivatives: the case of trans-1,2-C2H2F2

Infrared intensities of the trans-1,2-C2H2F2 molecule were computed using a comprehensive set of 52 basis sets combined with two DFT functionals (B3LYP, M06L) and two ab initio methods (MP2, CCSD). The aim was to identify the theoretical level that best reproduces the experimental intensity profile and to understand the systematic discrepancies previously reported, especially for the C–F stretching modes. Normalised relative intensities were employed to compare theoretical and experimental band patterns directly. The analysis revealed that the global mean absolute deviation (MAD) of relative intensities was minimised by specific double-zeta basis sets: M06L/Sapporo-DZP-2012-diffuse (MAD: 1.2 a.u.) and B3LYP/CBSB7 (MAD: 0.9 a.u.). Among ab initio methods, MP2 paired with the 6-311G(3d,3p) basis set yielded the lowest overall error (MAD: 0.5 a.u.). The study also demonstrates that anharmonic corrections do not significantly affect relative IR intensities for this system. Furthermore, the MP2/6-31G(d,p) level provides atomic polar tensor (APT) components that are effectively equivalent to those from larger basis sets within experimental uncertainty. The results highlight that balanced basis set quality and targeted error cancellation are more critical for accurate intensity prediction than sheer basis set size.