🔬 optics

Nanometer-resolution 3D Tomographic and Vectorial Near-field Imaging in Dielectric Optical Resonators

This paper presents a novel high-order sideband generation technique that achieves nanometer-resolution, three-dimensional tomographic, and vectorial near-field imaging within dielectric optical nano-resonators, overcoming previous experimental limitations in characterizing their internal electromagnetic wave distributions.

Original authors: Bingbing Zhu, Qingnan Cai, Yaxin Liu, Sheng Zhang, Weifeng Liu, Qiong He, Lei Zhou, Zhensheng Tao

Published 2026-07-30

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Bingbing Zhu, Qingnan Cai, Yaxin Liu, Sheng Zhang, Weifeng Liu, Qiong He, Lei Zhou, Zhensheng Tao

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine light not just as a beam you see, but as a bustling city of invisible waves. In the world of tiny machines called "nanophotonic structures," scientists build special neighborhoods out of glass and silicon to trap these waves. These neighborhoods are called "optical resonators." Think of them like a grand piano: when you hit a key, the sound waves bounce around inside the box, creating a rich, complex vibration. In these tiny silicon boxes, light does the same thing, creating intricate patterns of energy that can manipulate electricity, focus lenses, or even act as invisible shields.

For a long time, scientists could only peek at the surface of these vibrating light cities. It's like trying to understand a three-dimensional sculpture by only looking at its shadow on the wall. Tools that scan the surface are great, but they can't see deep inside because the "light waves" they use fade away almost instantly. Other tools that use electrons are even more shallow, like a diver who can only hold their breath for a few inches underwater. This leaves a huge mystery: what is the light actually doing deep inside the silicon? To build better future technologies, from super-fast computers to invisible cloaks, we need a map of the entire 3D interior, not just the surface.

This is where a team of researchers from Fudan University steps in with a clever new trick. They developed a method to take a "3D tomographic" picture of light deep inside a silicon resonator, achieving a resolution so sharp it can distinguish details as small as 130 nanometers deep and 920 nanometers wide. They didn't just take a picture; they mapped the light's direction (polarization) and its timing (phase), creating a complete, vectorial portrait of the invisible world.

Here is how they did it: Imagine you are in a dark room with a giant, silent drum (the silicon resonator). You want to know how the drumhead is vibrating deep inside, but you can't touch it. The researchers used two flashlights: a big, slow, red one (the "pump" laser) and a tiny, fast, blue one (the "probe" laser). When they shine these together, the silicon doesn't just reflect the light; it acts like a magical mixer. It takes the slow red waves and the fast blue waves and smashes them together to create entirely new colors of light, called "high-order sidebands."

Think of this like a DJ mixing two different songs. If you mix a slow beat with a fast beat, you might get a new, complex rhythm. In this case, the "new rhythm" is a beam of light that carries a secret code about exactly where the waves were inside the silicon when they mixed. The magic is that different "mixes" (different harmonic orders) are sensitive to different depths. One mix might only "hear" the vibrations near the top of the silicon, while another mix only "hears" the vibrations near the bottom. By listening to all these different mixes at once, the scientists can piece together a 3D puzzle, reconstructing the full shape of the light waves inside the material.

The paper shows that this technique works beautifully on a silicon cylinder about 1.5 micrometers thick. They found that the light inside wasn't just a simple blob; it had complex shapes, with some areas having no light at all (nodes) and others having strong fringes. They proved that their method could tell the difference between light vibrating up-and-down versus side-to-side (polarization) and could even detect if the light waves were shifting their timing by half a cycle (phase).

Crucially, the researchers demonstrated that this isn't just a simulation or a guess. They compared their experimental "photos" with computer simulations (using a method called Finite-Element-Method) and found they matched perfectly. They also showed that by using a special trick called "quantum-path interference," they could measure the phase of the light with incredible precision, seeing how the light waves shifted as they moved across the cylinder.

The paper argues against the idea that we are stuck with surface-level imaging. It explicitly rules out the limitations of traditional scanning microscopes, which can't see deeper than about 100 nanometers, and electron-based tools, which are limited to less than 50 nanometers. Instead, it suggests that by using these high-order nonlinear mixing processes, we can see much deeper, provided we have the right "mixing" orders to act as our depth sensors.

While the paper suggests this could be applied to other materials like metamaterials or even different types of light waves (like those in topological edge states), it stops short of claiming this is a finished, commercial product. Instead, it presents a powerful new tool that can be used to characterize these devices. The authors note that for even thicker or more complex materials, the method might need more "mixing orders" to stay accurate, and that current equipment has limits on how clearly it can see the deepest, faintest signals. However, for the silicon resonators they tested, the method worked with high confidence, offering a clear, 3D, vectorial map of light that was previously impossible to see.

In short, this paper hands us a new pair of glasses that lets us see the hidden, 3D dance of light inside solid materials. It turns a flat, blurry shadow into a sharp, colorful, and fully dimensional movie, opening the door to understanding and designing the next generation of light-based technology.

Technical Summary: Nanometer-Resolution 3D Tomographic and Vectorial Near-field Imaging in Dielectric Optical Resonators

Problem Statement
All-dielectric optical nano-resonators are critical components for manipulating electromagnetic waves and enhancing light-matter interactions due to their low losses and versatile responses. However, a significant experimental bottleneck remains: the inability to achieve full three-dimensional (3D) characterization of near-field distributions within these thick dielectric structures. Existing techniques face severe depth limitations. Scanning near-field optical microscopy (SNOM) offers sub-10 nm transverse resolution but is restricted to a probing depth of approximately 100 nm due to the rapid decay of evanescent fields. Similarly, electron-based techniques provide high spatial resolution but are limited to depths of less than 50 nm. Consequently, the complex near-field wave patterns deep inside micrometer-thick dielectric resonators, such as those supporting anapole modes, have remained largely inaccessible to experimental verification.

Methodology
To overcome these depth limitations, the authors developed a novel imaging technique based on High-Order Sideband Generation (HSG) microscopy. The method utilizes a nonlinear optical wave-mixing process between two laser fields:

A mid-infrared (MIR) pump laser ( $\lambda_{pu} \approx 13.4$ $\mu$ m) that resonantly excites near-fields within a silicon optical resonator.
A near-infrared (NIR) probe laser ( $\lambda_{pr} \approx 967$ nm) that is far off-resonance.

The interaction generates an output beam at wavelengths defined by $\lambda_{(m,n)} = \frac{\lambda_{pu}\lambda_{pr}}{m\lambda_{pu} + n\lambda_{pr}}$ , where $m$ and $n$ represent the number of probe and pump photons involved, respectively. By varying the harmonic orders ( $m, n$ ), the system generates a broad spectrum spanning more than an octave.

The core innovation lies in exploiting the phase-sensitivity of these different harmonic orders. The longitudinal resolution is governed by the variation in the longitudinal wavevector mismatch ( $\delta k_z$ ), which depends on the spectral bandwidth of the output beam and the probe laser. By employing multiple high-order processes (HSG orders), the authors create distinct "gate functions" ( $g_{(m,n)}(z)$ ) that exhibit unique depth-dependent sensitivities. This allows the system to probe near-fields at different depths within the material.

The reconstruction process involves:

Tomographic Reconstruction: Modeling the HSG intensity as a coherent integration of nonlinear polarizations along the depth ( $z$ ). A global optimization algorithm minimizes the difference between experimental images and modeled images to retrieve complex coefficients describing the near-field distribution $E_{NF}(x, y, z)$ .
Vectorial Imaging: By controlling the polarization of the probe beam and the detection polarizer, the method resolves specific vector components ( $E_x, E_y$ ) of the near-field.
Phase Resolution: The technique utilizes Quantum-Path Interference (QPI) between neighboring HSG orders (achieved by broadening the probe pulse spectrum). The interference patterns in the spectral domain encode the local phase of the near-field, allowing for phase retrieval.

Key Results
The authors demonstrated this technique on micrometer-thick silicon cylindrical resonators supporting mid-infrared anapole modes:

3D Tomography: Using six different HSG orders, the team reconstructed the 3D near-field wave patterns inside a 1.5 $\mu$ m thick silicon resonator. The experimental reconstructions showed excellent quantitative agreement with Finite-Element Method (FEM) simulations.
Resolution: The method achieved a transverse spatial resolution of ~920 nm (sub-micrometer) and a conditional longitudinal spatial resolution of ~130 nm. The longitudinal resolution is described as "conditional," relying on the signal-to-noise ratio and the complexity of the field distribution.
Vectorial Capability: The system successfully resolved distinct spatial distributions for the $x$ - and $y$ -polarized near-field components, achieving a polarization extinction ratio of ~30 dB for second-harmonic HSG orders.
Phase Mapping: Through QPI, the authors mapped the phase of the near-field, observing $\pi$ phase shifts across specific trajectories within the resonator, which matched FEM predictions.

Significance and Claims
The paper claims to present the first nanometer-resolution, 3D tomographic, and vectorial imaging of optical near-field wave patterns within micrometer-thick dielectric resonators. The significance of this work is threefold:

Overcoming Depth Limits: It provides a solution to the depth limitation inherent in SNOM and electron-based microscopy, enabling the characterization of fields deep inside dielectric materials.
Comprehensive Characterization: By simultaneously providing spatial, vectorial (polarization), and phase information, the method offers a complete picture of near-field phenomena, which is essential for verifying novel nanophotonic concepts like anapole modes.
Broad Applicability: The authors note that since inherent optical nonlinearity is abundant in dielectrics and semiconductors, this approach is potentially applicable to a wide range of dielectric metamaterials. They suggest it could be adapted for other wave phenomena, such as hyperbolic phonon-polaritons and topological photonic edge states, to image their 3D distribution and propagation.

The authors maintain a modest tone regarding the current limitations, noting that the high longitudinal resolution is conditional on the signal-to-noise ratio and that more complex field distributions would require additional HSG orders or more sophisticated reconstruction algorithms. They also acknowledge that while the current setup uses MIR pumps, the concept could be adapted for shorter wavelengths, though this presents detection challenges due to the shift of harmonic energy into the extreme ultraviolet range.

More like this