Scattering scanning near-field optical microscopy (s-SNOM) has emerged as a powerful imaging and spectroscopic tool for investigating nanoscale heterogeneities in biology, quantum matter, and electronic and photonic devices. However, many materials are defined by a wide range of fundamental molecular and quantum states at far-infrared (FIR) resonant frequencies currently not accessible by s-SNOM. Here we show ultrabroadband FIR s-SNOM nanoimaging and spectroscopy by combining synchrotron infrared radiation with a novel fast and low-noise copper-doped germanium (Ge:Cu) photoconductive detector. This approach of FIR synchrotron infrared nanospectroscopy (SINS) extends the wavelength range of s-SNOM to 31 μm (320 cm-1, 9.7 THz), exceeding conventional limits by an octave to lower energies. We demonstrate this new nanospectroscopic window by measuring elementary excitations of exemplary functional materials, including surface phonon polariton waves and optical phonons in oxides and layered ultrathin van der Waals materials, skeletal and conformational vibrations in molecular systems, and the highly tunable plasmonic response of graphene.

We measured the near-field infrared response of MnPS3 in bulk, few-, and single-layer form and compared our findings with traditional far-field vibrational spectroscopies, a symmetry analysis, and first-principles lattice dynamics calculations. Trends in the Bu mode near 450cm-1 are striking, with the disappearance of this structure in the thinnest sheets. Combined with the amplified response of the activated Ag mode and analysis of the Au+Bu features, we find that symmetry is unexpectedly increased in few- and single-sheet MnPS3 due to a restoration of the threefold axes of rotation. Monoclinicity in this system is therefore a consequence of the long-range stacking pattern and temperature rather than local structure.

We combine synchrotron-based near-field infrared spectroscopy and first principles lattice dynamics calculations to explore the vibrational response of CrPS4 in bulk, few-, and single-layer form. Analysis of the mode pattern reveals a C2 polar + chiral space group, no symmetry crossover as a function of layer number, and a series of non-monotonic frequency shifts in which modes with significant intralayer character harden on approach to the ultra-thin limit whereas those containing interlayer motion or more complicated displacement patterns soften and show inflection points or steps. This is different from MnPS3 where phonons shift as 1/size2 and are sensitive to the three-fold rotation about the metal center that drives the symmetry crossover. We discuss these differences as well as implications for properties such as electric polarization in terms of presence or absence of the P-P dimer and other aspects of local structure, sheet density, and size of the van der Waals gap.

Modern scattering-type scanning near-field optical microscopy (s-SNOM) has become an indispensable tool in material research. However, as the s-SNOM technique marches into the far-infrared (IR) and terahertz (THz) regimes, emerging experiments sometimes produce puzzling results. For example, "anomalies" in the near-field optical contrast have been widely reported. In this Letter, we systematically investigate a series of extreme subwavelength metallic nanostructures via s-SNOM near-field imaging in the GHz to THz frequency range. We find that the near-field material contrast is greatly impacted by the lateral size of the nanostructure, while the spatial resolution is practically independent of it. The contrast is also strongly affected by the connectivity of the metallic structures to a larger metallic "ground plane". The observed effect can be largely explained by a quasi-electrostatic analysis. We also compare the THz s-SNOM results to those of the mid-IR regime, where the size-dependence becomes significant only for smaller structures. Our results reveal that the quantitative analysis of the near-field optical material contrasts in the long-wavelength regime requires a careful assessment of the size and configuration of metallic (optically conductive) structures.

The strong electron-lattice interactions in correlated electron systems provide unique opportunities for altering the material properties with relative ease and flexibility. In this Rapid Communication, we use localized strain control via a focused-ion-beam patterning of TiO2 substrates to demonstrate that one can selectively engineer the insulator-to-metal transition temperature, the fractional component of the insulating and metallic phases, and the degree of optical anisotropy down to the length scales of the intrinsic phase separation in VO2 thin films without altering the quality of the films. The effects of localized strain control on the strongly correlated electron system are directly visualized by state-of-the-art IR near-field imaging and spectroscopy techniques and x-ray microdiffraction measurements.

The underlying physics behind an experimental observation often lacks a simple analytical description. This is especially the case for scanning probe microscopy techniques, where the interaction between the probe and the sample is nontrivial. Realistic modeling to include the exact details of the probe is widely acknowledged as a challenge. Due to various complexity constraints, the probe is often only approximated in a simplified geometry, leading to a source for modeling inconsistencies. On the other hand, a well-trained artificial neural network based on real data can grasp the hidden correlation between the signal and the sample properties, circumventing the explicit probe modeling process. In this work we show that, via a combination of model calculation and experimental data acquisition, a physics-infused hybrid neural network can predict the probe-sample interaction in the widely used scattering-type scanning near-field optical microscope. This hybrid network provides a long-sought solution for accurate extraction of material properties from tip-specific raw data. The methodology can be extended to other scanning probe microscopy techniques as well as other data-oriented physical problems in general.