🔬 applied physics

Near-Field Characterisation of Guided Modes in WS2 Nanobeams and Quasi-Bulk Crystals

This study utilizes hyperspectral cavity-enhanced imaging and scattering-type scanning near-field optical microscopy (s-SNOM) to characterize the wavelength-dependent modal confinement, attenuation, and dispersion of guided modes in WS2 nanobeams and quasi-bulk crystals, while establishing high-resolution bounds on the material's extinction coefficient and identifying critical measurement artifacts in nanoscale waveguides.

Original authors: Zara S. Taylor, Luke M. Hallacy, Xuerong Hu, Oliver T. Williams, Simone Strohmair, Fabian Felixberger, Alexander J. Knight, Timothy Chester-Parsons, Luke R. Wilson, Alexander I. Tartakovskii

Published 2026-07-02

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Zara S. Taylor, Luke M. Hallacy, Xuerong Hu, Oliver T. Williams, Simone Strohmair, Fabian Felixberger, Alexander J. Knight, Timothy Chester-Parsons, Luke R. Wilson, Alexander I. Tartakovskii

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 you are trying to build a super-fast, tiny city for light to travel through, similar to how we build cities for cars. In this city, the "roads" are called waveguides, and the "cars" are beams of light. For a long time, we've used silicon (like computer chips) to build these roads, but silicon has a few problems: it's hard to make it work with light of certain colors, and it doesn't guide light as tightly as we'd like.

Enter WS2 (Tungsten Disulfide). Think of WS2 as a new, super-special kind of "optical asphalt." It's made of incredibly thin, layered sheets (like a stack of paper) that have some magical properties: they can squeeze light into very tiny spaces and guide it with very little loss, especially for the colors of light our eyes can see and the near-infrared.

However, before we can build our light-city with this new material, we need to answer two big questions:

How "sticky" is the road? (Does the material absorb the light and stop it, or does it let it pass through?)
How well does the road guide the light? (Does the light stay on the road, or does it leak out?)

This paper is like a team of detectives using high-tech flashlights to answer these questions.

The Mystery of the "Sticky" Road (Absorption)

Scientists knew WS2 was good, but they weren't sure exactly how much light it "ate" (absorbed) when the light wasn't quite the right color to make the material glow. Different previous studies gave wildly different answers—some said it was almost invisible to light, others said it was quite sticky.

To solve this, the researchers used a special tool called Cavity-Enhanced Imaging. Imagine trying to hear a whisper in a noisy room. If you put the whisperer in a small, echoey box (a cavity), the sound bounces around and gets louder, making it easier to hear. They did this with light. By trapping light in a tiny box with a piece of WS2 inside, they could measure exactly how much light the material absorbed.

The Verdict: They found that WS2 is indeed very transparent (low absorption) in the visible and near-infrared range, but it's not perfectly clear. It's about as "sticky" as silicon is for these colors. This means it's great for tiny, short-distance light circuits (microns), but maybe not for massive, long-distance light highways (wafer-scale) just yet.

The Flashlight Detective (s-SNOM)

To see how light travels inside these tiny WS2 roads, the team used a tool called s-SNOM.

Imagine you are trying to see the ripples in a pond, but the pond is so small you can't see it with your eyes. You take a very sharp, vibrating needle (like a tiny tuning fork) and gently tap the surface. As the needle moves, it creates a tiny splash of light that acts like a flashlight, illuminating the ripples right under the tip.

In this experiment, the "needle" is a metal tip vibrating on the WS2 crystal. When the laser hits the tip, it creates a tiny burst of light that couples into the WS2 road. The light travels down the road, hits the edge, and bounces back. The needle detects the interference between the light coming directly from the tip and the light bouncing back from the edge. This creates a pattern of "fringes" (like the stripes on a zebra) that tells the scientists exactly how the light is moving.

What They Found on the Roads

They looked at two types of "lanes" on the road:

The TE Lane: Light waves vibrating side-to-side.
The TM Lane: Light waves vibrating up-and-down.

The TE Lane (Side-to-Side):
This lane is very good at keeping light confined, but as the color of the light changes (gets redder), the light starts to "leak" out more. The researchers found a direct link: the more the light leaks out of the road, the more it gets scattered by tiny bumps on the road surface, causing it to die out faster. It's like a car driving on a bumpy road; if the car is already low to the ground (less confined), the bumps knock it off course faster.

The TM Lane (Up-and-Down):
This lane is trickier. In very thin roads, this type of light doesn't stay on the road at all; it leaks straight down into the ground (the substrate). However, in thicker sections, it stays put. The researchers found that this lane behaves differently than the TE lane, becoming less lossy as the light gets redder because it stops interacting with the surface bumps so much.

The "Narrow Road" Problem (Nanobeams)

The team also tested WS2 roads that were carved into tiny beams, only a few hundred nanometers wide (thinner than a human hair).

Here, they discovered a sneaky problem. When the road is so narrow that it's almost the same size as the light wave itself, the light doesn't just travel straight down the middle. It creates a complex interference pattern, like ripples bouncing off the side walls of a narrow canyon.

If you try to measure the light speed in the middle of the road, you might get the wrong answer because the light is actually bouncing back and forth across the width of the road. The researchers found that if you don't look at the whole width of the road, you might think the light is moving slower than it really is. It's like trying to measure the speed of a car by only looking at the left tire; if the car is swerving, your measurement will be wrong.

The Bottom Line

This paper is a "quality control" report for WS2 as a material for future light-based computers.

Good News: WS2 is a fantastic material for guiding light in tiny, integrated circuits. It has low loss and can squeeze light into very small spaces.
Bad News: It's not perfect. It absorbs a bit more light than we'd like for very long distances, and measuring it is tricky because the light behaves strangely in tiny, narrow structures.
The Tool: They proved that the "flashlight needle" (s-SNOM) is a great way to check the quality of these light roads, as long as you know how to interpret the patterns it sees.

In short, WS2 is a promising new material for the next generation of tiny, fast optical chips, but engineers need to be very careful about how they design the roads and how they measure the traffic to avoid getting lost in the details.

Technical Summary: Near-Field Characterisation of Guided Modes in WS2 Nanobeams and Quasi-Bulk Crystals

Problem Statement
Tungsten disulfide (WS2) is a promising material for next-generation photonic integrated circuits (PICs) due to its exceptionally high in-plane refractive index, strong optical anisotropy, and compatibility with heterogeneous integration via van der Waals forces. However, significant ambiguities remain regarding its waveguiding performance. Specifically, reported values for the sub-bandgap extinction coefficient ( $\kappa$ ) vary by orders of magnitude, leading to large uncertainties in predicted modal decay lengths and the feasibility of wafer-scale integration. Furthermore, while the Transverse Electric (TE) modes are well-studied, the behavior of Transverse Magnetic (TM) modes and higher-order modes in highly anisotropic WS2 waveguides remains under-characterized. Existing studies often lack systematic, wavelength-dependent analysis of modal confinement and attenuation, and the utility of scattering-type scanning near-field optical microscopy (s-SNOM) as a diagnostic tool for these anisotropic structures has not been fully established.

Methodology
The authors employed a multi-faceted experimental approach to resolve these ambiguities:

Hyperspectral Cavity-Enhanced Absorption Spectroscopy (iCEAS): Using a "Qlibri Nano" cavity microscope, the team measured the absorption spectrum of WS2 within the 872–934 nm range. This technique provided high-sensitivity, model-independent quantification of the extinction coefficient ( $\kappa$ ) in both weakly and strongly absorbing regions of quasi-bulk crystals.
Scattering-type Scanning Near-Field Optical Microscopy (s-SNOM): The authors utilized s-SNOM to probe guided modes (TE $_0$ $_{0}$ , TM $_0$ $_{0}$ , and higher-order modes) in both quasi-bulk WS2 crystals and fabricated nanobeams across the 800–1400 nm spectral range.
- Quasi-Bulk Samples: Large, pristine crystals (>100 $\mu$ m) with minimal surface roughness were used to establish baseline confinement and loss characteristics, minimizing scattering from sidewalls.
- Nanobeam Samples: Arrays of 200 nm-thick WS2 nanobeams with widths ranging from 400 nm to 650 nm were fabricated via Electron Beam Lithography (EBL) and Reactive Ion Etching (RIE) to simulate PIC-relevant geometries.
Data Analysis: Near-field fringe patterns were analyzed via Fourier Transform. The resulting spectra were fitted using square-root Lorentzian functions to extract effective refractive indices ( $n_{eff}$ ), decay constants ( $\gamma$ ), and amplitudes. Theoretical simulations using Lumerical FDE and FDTD were performed to compare experimental results with expected modal behavior.

Key Results

Extinction Coefficient Bounds: The iCEAS measurements established high-resolution upper and lower bounds for the WS2 extinction coefficient in the visible/NIR edge. The results indicate an average absorption loss of $23 \pm 18$ dB/mm in weakly absorbing regions and up to $118 \pm 85$ dB/mm in strongly absorbing areas. These values suggest that while WS2 is adequate for micron-scale waveguiding, its intrinsic loss may be too high for wafer-scale integration in this wavelength regime.
Modal Dispersion and Confinement:
- TE $_0$ Mode: Exhibits high dispersion ( $\Delta n_{eff} \approx -0.98$ from 800–1300 nm) due to the highly dispersive in-plane refractive index. The decay constant increases with wavelength, correlating with a decrease in the filling factor (field confinement), indicating that loss is driven by increased field overlap with surface defects.
- TM $_0$ Mode: Shows reduced modal dispersion ( $\Delta n_{eff} \approx -0.29$ ) as it interacts primarily with the less dispersive out-of-plane index. However, it approaches cutoff at longer wavelengths, leading to significant leakage into the substrate that cause it to exhibit a negative trend in loss with wavelength, directly opposing the trends observed for the TE $_0$ mode.
s-SNOM as a Diagnostic Tool: While absolute loss values derived from s-SNOM are inflated by measurement geometry (estimated at an additional ~100–140 dB/mm due to coupling and scattering effects), the technique reliably captures relative modal trends. The study confirms that s-SNOM can effectively benchmark waveguide performance and identify mode-specific loss mechanisms.
Nanobeam Artifacts: In sub-wavelength nanobeams, the authors identified significant artifacts arising from transverse interference and spatial sampling. When the structure size approaches the excitation wavelength, fringe patterns vary significantly along the short axis of the beam. This leads to shifts in extracted effective indices (up to $\Delta n_{eff} = -0.25$ ) if not properly accounted for, highlighting the need for position-resolved analysis rather than simple line-profile averaging.

Significance and Claims
The paper claims to resolve long-standing ambiguities regarding the optical loss and modal behavior of WS2 waveguides. By combining iCEAS and s-SNOM, the work provides the first systematic, wavelength-resolved characterization of both TE and TM modes in WS2.

The authors assert that their findings demonstrate s-SNOM as a powerful, albeit geometry-dependent, tool for diagnosing waveguide performance in highly anisotropic van der Waals materials. They conclude that while absolute loss extraction requires careful calibration against reference structures, s-SNOM is capable of providing direct feedback for waveguide design, fabrication optimization, and material selection. Furthermore, the identification of transverse interference artifacts in nanobeams serves as a critical methodological correction for future studies of sub-wavelength photonic structures, emphasizing that spatial sampling effects can drastically alter extracted modal properties.