Commissioning-oriented AFM-TERS workflow on planar gold thin films using benchmark dyes and near-/far-field...

Tip-Enhanced Raman Spectroscopy (TERS) provides nanoscale chemical sensitivity, but its routine use remains constrained by thermal drift, unstable optical coupling, heterogeneous hotspot formation, and inconsistent approaches to signal quantification. Here, we present a commissioning-oriented AFM-TERS workflow for planar Au thin films under the specific instrumental and substrate conditions used in this study. A 20 min thermal stabilization period reduced the lateral drift rate by approximately 3.6× (from 186 to 51 nm·min^-1), thereby improving platform stability; however, the residual drift remains significant for long spectral maps and requires cautious interpretation of pixel-level co-localization. Optical coupling was standardized through a camera-based scattering-footprint analysis used as an operational alignment proxy, not as a direct measurement of the TERS hotspot or as a quantitative predictor of absolute Raman enhancement. Near-field mapping of Rhodamine 6G, Methylene Blue, and Crystal Violet confirmed local molecular identification at selected nanoscale positions, while paired far-field controls were used to evaluate the near-field contribution. Under the present experimental conditions, ensemble SERS on planar Au was practically limited to ≈10^-6 mol·L^-1, whereas TERS allowed local identification of R6G at selected positions on samples prepared down to 10^-8 mol·L^-1 via isolated, hotspot-mediated observations rather than as evidence of a practical detection capability. These measurements should not be interpreted as a formal statistical detection-limit study. Concentration-dependent measurements revealed strong local variability, including regimes in which substrate-mediated far-field hotspots locally dominated the response (FF > NF). We therefore recommend reporting paired near-field/far-field (NF/FF) intensity ratios and the differential signal metric (ΔI) as directly measurable descriptors that complement, rather than replace, conventional enhancement-factor estimates when the assumptions required for molecule-normalized EF calculations are uncertain.