The concept of utilizing a near-field technique to circumvent the diffraction limit dates back to the early 20th century, proposed by Edward Synge^[1]. After the first demonstration of this idea, in 1972 in microwaves^[2] and 1984 in visible light^[3,4], various practices with similar near-field approaches emerged. Hillenbrand et al.^[5] recently reviewed scattering-type scanning near-field optical microscopy (s-SNOM) as a specific type of near-field scanning technique. The detection limit of this near-field scanning technique is characterized by the sharpness of the metallic tip, rather than the wavelength of the electromagnetic wave. Thus, the potential to scan with a very broad frequency range while maintaining an extremely high spatial resolution (nominally 10–100 nm) makes this proposal extremely promising. By analyzing the local dielectric constant ϵ(x) as a function of scanning frequency and controlling other environmental parameters, a wealth of physical information can be extracted.

Hillenbrand et al. first introduced the mechanism of s-SNOM [Fig. 1(A)]. A Gaussian beam of an electromagnetic field (in THz, IR, or visible frequencies) is focused onto the surface of a sample. The size of the beam is limited by the diffraction limit and is thus frequency-dependent. However, in the s-SNOM setup, a tip is placed directly above the focal spot on the surface of the material. In the vicinity of the tip, the electromagnetic field is strongly influenced by near-field interaction and significantly enhanced^[6,7]. The scattered electromagnetic field, which can then be collected, has both near-field and far-field contributions, where only the near-field part encodes information on the local property of the material with a resolution close to the tip size. The amplitude and the relative phase of the scattered field can then be detected, which provides crucial information on the local dielectric constant^[8]. Hillenbrand et al. discussed different models and numerical simulations on reconstructing the local dielectric constant out of the detected scattered signal. Different pathways of light-matter coupling involving charge carriers^[9,10], phonons^[11,12], and molecular vibrations^[13,14] are to be considered. The authors also reviewed the methods to reconstruct a local dielectric constant in the presence of irregular probe and sample geometries, including analytical modeling^[15–17], numerical simulations^[18], and machine learning approaches^[19,20].

Hillenbrand et al. then compared the s-SNOM technique with other existing near-field scanning techniques (such as aperture scanning near-field optical microscopy, or other existing probes like microwave impedance microscopy). The incident beam in s-SNOM is apertureless (propagated through free space) and thus escapes the technological difficulty of fabricating tips (such as building a tiny hole with a metallic coating to guide the electromagnetic wave). s-SNOM works with standard metal-coated AFM tips. It also does not require impedance matching. The trade-off is a strong far-field contribution to the scattered light. They mention that researchers usually make the tip oscillate vertically with a small amplitude at frequency Ω and look for a higher-harmonic contribution (nΩ) on the scattered light^[15,21]. Since the far-field scattering process has smaller nonlinearity than the near-field signal, filtering the higher-harmonic component of n≥3 usually isolates well the near-field scattered field^[8,22,23].

After addressing the basic mechanism of s-SNOM, the authors reviewed the setup of the experimental system and introduced different working modes of s-SNOM. The s-SNOM setup is highly versatile and can accommodate a broad spectrum of measurement conditions. It can work with different laser sources: monochromatic or broadband, continuous wave or pulsed. When s-SNOM has a monochromatic input, the amplitude-resolved and phase-resolved maps can be obtained in a coherent Michelson interferometer setup^[8]. A high-resolution microscopy map can be detected at a certain frequency. Furthermore, with a tunable monochromatic source, such as Ti:sapphire lasers (visible to near-infrared), quantum cascade lasers (mid-infrared), or free-electron lasers, a high-resolution spectroscopic map can be obtained^[24–26]. The s-SNOM can also work with broadband sources^[27], where the setup works as a nano-Fourier transform (FT) spectrometer, acquiring a wider range of spectroscopic data in a shorter time while lowering the signal-to-noise ratio (SNR) as a trade-off.

The authors also discussed the working environment of s-SNOM. The versatility of s-SNOM makes it a powerful platform for investigating materials under a broad range of external conditions [Fig. 1(B)]. Measurements can be performed at both elevated^[28] and cryogenic temperatures^[29,30], enabling direct observation of nanoscale dielectric or conductivity changes across phase transitions such as the metal-to-insulator transitions in VO2 and V2O3^[31,32]. Furthermore, by employing ultrafast pump-probe schemes with tunable delay times between femtosecond pulses, s-SNOM provides access to transient dynamics and nonequilibrium phenomena at the nanoscale^[33–35]. Beyond temperature and temporal control, s-SNOM has been successfully integrated with tunable electric^[36,37] and magnetic field^[38] environments, facilitating the study of field-induced transitions and symmetry-breaking effects. Finally, strain engineering, either statically or dynamically, offers another powerful degree of freedom, allowing s-SNOM to reveal strain-mediated changes in structural and electronic properties^[39]. These capabilities highlight the adaptability of s-SNOM for probing emergent quantum and correlated phenomena in low-dimensional systems.

Hillenbrand et al. summarized the impact of s-SNOM in three major arenas [Fig. 1(C)]. (i) Property mapping: it can map out and provide insights into the spatial inhomogeneities in a wide variety of systems, like phase separation in block copolymers^[40], coexisting phases in electrode materials for Li-ion batteries^[41], carrier density and domain structures across semiconductors, Mott and topological insulators, and perovskites^[9]. (ii) Structure and device characterization: it provides label-free images of viruses, living cells under liquid, core-shell polymers, 2D moire superlattices, and plasmonic/polaritonic metasurfaces^[42–45]. (iii) Phenomena exploration: complex phenomena happen in all types of systems and are hard to track. s-SNOM can track the metal to insulator transition processes, heterogeneous catalytic and enzymatic processes, and physics processes like graphene plasmon propagation, where the tip can provide momentum matching to launch or detect ultra-confined plasmons, phonons, and exciton polaritons^[37,46–48].

The authors then discussed the perspectives of this technique. s-SNOM will make major leaps forward as instrumentation, light sources, scanning modalities, environmental control, and AI-driven analysis all improve. For instance, synchrotron sources, free-electron lasers, custom-shaped tips, and quantum-cascade lasers are expected to make far-infrared and terahertz nano-FTIR a routine tool, while compact optical parametric oscillation (OPO) lasers and electro-optic sampling will enhance its time resolution greatly. Integrating s-SNOM with other microscopies is another routine to gather optical, mechanical, electronic, and magnetic data simultaneously. New cryogenic, high-field, strain-tunable, and liquid-cell stages will make it possible to reveal fragile quantum phases of interest, and study functional devices and living cells at roughly 10-nm resolution. Machine learning and more accurate modeling will clean noisy hyperspectral data, speed up imaging, and bring true 3D nano-tomography within reach. These advances will enable hard tasks, like real-time views of electrolyte breakdown and interphase growth in batteries, material degradation and polymer aging, detection of micro- and nanoplastics, mapping of protein structures linked to neurodegenerative diseases, and probing millielectron-volt excitations in topological semimetals and superconductors.

To summarize, from the early 20th century to now, s-SNOM has evolved from a niche microscopy tool into a versatile, powerful technique for chemistry, biology, and quantum materials. Continued innovations in light sources, detection mechanisms, and data analysis will likely push its spatial and spectral resolutions to new limits, enable operando studies under extreme conditions, and illuminate emergent quantum phenomena that are inaccessible to far-field optics. Echoing Phil Anderson’s dictum that “more is different”^[49], the growing capabilities of s-SNOM promise transformative discoveries across the breadth of science.