Machine Learning for Optical Scanning Probe Nanoscopy

Abstract

The ability to perform nanometer‐scale optical imaging and spectroscopy is key to deciphering the low‐energy effects in quantum materials, as well as vibrational fingerprints in planetary and extraterrestrial particles, catalytic substances, and aqueous biological samples. These tasks can be accomplished by the scattering‐type scanning near‐field optical microscopy (s‐SNOM) technique that has recently spread to many research fields and enabled notable discoveries. Herein, it is shown that the s‐SNOM, together with scanning probe research in general, can benefit in many ways from artificial‐intelligence (AI) and machine‐learning (ML) algorithms. Augmented with AI‐ and ML‐enhanced data acquisition and analysis, scanning probe optical nanoscopy is poised to become more efficient, accurate, and intelligent.

Near‐ and Far‐Field Observation of Phonon Polaritons in Wafer‐Scale Multilayer Hexagonal Boron Nitride Prepared by Chemical Vapor Deposition

“…Thus, the analysis of PhP interference patterns may become a valuable method of quality assessment of CVD-grown 2D materials and for studying growth defects. Artificial intelligence may be applied to analyze the s-SNOM images, [39][40][41] based on a training of a neural network with simulated polariton interference patterns (according to the model presented in this work) obtained for different defect distributions. Despite strong PhP scattering, our ML-hBN allows the fabrication of Fabry-Pérot phonon polariton resonators with quality factors up to Q ≈ 50, demonstrating the potential of ML-hBN for large-scale fabrication of PhP nanoresonators for infrared gas sensing [36] and phonon-polariton-assisted detection of infrared radiation.…”

Ultrafast anisotropic dynamics of hyperbolic nanolight pulse propagation

“…The time-dependent evolution of polariton wave vectors and associated topological transitions observed in our study offers exciting opportunities for dynamically tailoring the local density of photonic states in various applications ( 19 ) and exploring high-dimensional time-varying optics ( 43 , 44 ). We anticipate that the use of artificial intelligence techniques ( 45 ) and machine learning algorithms ( 46 ) will become increasingly important to accelerate the acquisition of high-dimensional spatiotemporal data and improve the imaging quality and accuracy. The spatiotemporal dynamics of HP pulses demonstrated here are essential for developing anisotropy-driven ultrafast nanophotonic and thermal applications.…”

Accelerated Nano-Optical Imaging through Sparse Sampling

“…However, the nature of sampling a dense evenly spaced grid with the sufficient integration time required to reach a reasonable signal-to-noise ratio results in long scan times. To overcome speed limitations of common raster-based SPM, sparse sampling supplemented by reconstruction methods has emerged as a viable improvement. − Likewise, one can expect sparse sampling to excel in SNOM, especially when measuring periodic features such as polaritonic fringes. Furthermore, implementation of sparse sampling is a critical component of more complex data acquisition methods that can dynamically choose the scan path to optimize the scan information collected .…”