Thickness-matched FT-IR benchmarking reveals spectral bias in AFM-IR measurements of supported silica films.

Atomic force microscopy-infrared (AFM-IR) spectroscopy enables nanoscale chemical characterization, but apparent peak shifts in supported thin films can combine genuine film-structure evolution with thickness-dependent measurement response. Here, replicated AFM-IR spectra of TEOS-derived silica films with nominal thicknesses of 10 nm, 50 nm, 100 nm and 200 nm were benchmarked against thickness-matched transmission Fourier-transform infrared (FT-IR) spectra. Film thicknesses were determined by X-ray reflectivity. The principal Si-O-Si band shifted to lower wavenumber with increasing thickness in both methods, but the total shift was approximately 23 cm-1 in AFM-IR and only 6 cm-1 in FT-IR. The AFM-IR thickness dependence therefore substantially exceeded the trend shared by the two methods, indicating additional technique-specific spectral weighting. A high-wavenumber area fraction defined without component fitting showed the largest excess in AFM-IR relative to FT-IR for the 10 nm film, which also exhibited the largest between-method peak-position difference. The AFM-IR/FT-IR area-fraction ratio decreased from 2.07 at 10 nm to 1.06-1.08 at 100 nm to 200 nm, and this thickness trend was retained when the baseline and integration windows were varied. FT-IR/XRR-derived SiOH number-density trends were also consistent with an AFM-IR-resolved SiOH-related O-H feature near 3620 cm-1 in the 50 nm, 100 nm and 200 nm films. Thickness-matched far-field benchmarking combined with model-independent line-shape analysis provides a practical framework for evaluating the measurement response embedded in AFM-IR peak changes before attributing them to nanoscale chemical heterogeneity.