Nanoscale bond-selective imaging by computational fusion of atomic force microscopy and coherent anti-Stokes Raman scattering microscopy.

Vibrational microscopy based on coherent Raman scattering is a powerful tool for high-speed chemical imaging, but its lateral resolution is bound to the optical diffraction limit. On the other hand, atomic force microscopy (AFM) provides nano-scale spatial resolution, yet with lower chemical specificity. In this study, we leverage a computational approach called pan-sharpening to merge AFM topography images and coherent anti-Stokes Raman scattering (CARS) images. The hybrid system combines the advantages of both modalities, providing informative chemical mapping with ∼20 nm spatial resolution. CARS and AFM images were sequentially acquired on a single multimodal platform, which facilitates image co-localization. Our image fusion approach allowed for discerning merged neighboring features previously invisible due to the diffraction limit and identifying subtle unobservable structures with the input from AFM images. Compared to tip-enhanced CARS measurement, sequential acquisition of CARS and AFM images enables higher laser power to be used and avoids any tip damage caused by the incident laser beams, resulting in a significantly improved CARS image quality. Together, our work suggests a new direction for achieving super-resolution coherent Raman scattering imaging of materials through a computational approach.