Vectorial engineering of second-harmonic generation in silicon-based waveguides integrated with 2D materials

This paper demonstrates that accounting for the full vectorial and tensorial nature of electromagnetic fields and second-order susceptibility in silicon nitride waveguides integrated with monolayer MoS$_2$ enables efficient, phase-matched cross-polarized second-harmonic generation, achieving a 220-fold enhancement over free-space excitation and establishing fundamental design guidelines for 2D-material-based nonlinear photonic devices.

The Gist

Here is an explanation of the paper using simple language and creative analogies.

The Big Picture: Turning Light into New Colors

Imagine you have a laser beam that is deep red (like a ruby). You want to turn it into bright green light. In the world of physics, this is called Second-Harmonic Generation (SHG). It's like taking two low-energy red photons and smashing them together to create one high-energy green photon.

Usually, to do this efficiently, you need special crystals. But silicon—the material used to make computer chips—is terrible at this. It's like trying to mix paint with a spoon made of glass; the silicon just doesn't have the right "structure" to let the colors mix.

The Problem: Silicon is Too Symmetrical

Think of a silicon crystal like a perfectly symmetrical snowflake. If you flip it over, it looks exactly the same. Because of this perfect symmetry, it cancels out the ability to mix light frequencies. To get silicon to do this, scientists usually have to stretch it, apply electric fields, or glue other complex crystals to it. These methods are like trying to fix a leaky pipe with duct tape and a hammer: they work, but they are messy, hard to build, and hard to scale up.

The Solution: The 2D "Sticker"

The researchers in this paper found a clever workaround. Instead of trying to change the silicon, they stuck a tiny, ultra-thin "sticker" on top of it. This sticker is made of Molybdenum Disulfide (MoS₂), a material that is only one atom thick (a "monolayer").

Think of MoS₂ as a magic trampoline. Unlike the rigid, symmetrical silicon, this trampoline is lopsided (it lacks inversion symmetry). When light hits it, it bounces back in a way that allows the colors to mix. Because it's so thin, you can easily stick it onto a silicon chip without needing to match the microscopic "grid" of the silicon (a process called lattice matching, which is usually very difficult).

The Surprise: It's Not Just About "Up and Down"

Here is where the paper gets really interesting.

For a long time, scientists used a simple model to predict how this would work. They thought: "If the main part of the light wave is pointing straight up (perpendicular to the sticker), and the sticker is flat, they won't touch. No mixing!"

It's like trying to rub a coin against a table. If you hold the coin flat against the table, it rubs. If you hold it on its edge (perpendicular), it doesn't touch the surface much.

The paper proves this simple model is wrong.

The researchers discovered that even if the main part of the light is pointing "up" (orthogonal to the sticker), the light still has tiny "whiskers" or side-components that wiggle along the direction of the waveguide. The MoS₂ sticker grabs onto these "whiskers" and still manages to mix the colors.

The Analogy: Imagine trying to push a swing. The simple model says, "If you push straight down on the seat, the swing won't move." But the researchers found that if you push the chain of the swing (the side component), you can still make it swing! They realized that the "side wiggles" of the light are just as important as the main push.

The Breakthrough: Tuning the Waveguide

Once they understood that these "side wiggles" mattered, they decided to build a custom silicon track (a waveguide) to make the process super efficient.

1. Phase Matching: Imagine two runners on a track. If one runs fast and the other runs slow, they will drift apart, and they can't high-five. "Phase matching" is like adjusting the track so both runners (the incoming red light and the outgoing green light) run at the exact same speed.
2. The Result: By designing a specific width for the silicon track, they made the red light and the green light run in perfect sync.

The Results: A Massive Boost

The team tested their new design:

• Without the sticker: The silicon chip produced almost no green light.
• With the sticker (but no tuning): They got a little bit more green light (about 2x to 3x better).
• With the sticker AND the tuning (Phase Matching): They got a 220 times boost in efficiency compared to shining a laser directly at a loose piece of MoS₂ in the air.

Why This Matters

This paper is a game-changer for two reasons:

1. It's a Blueprint: They showed that you don't need to ignore the "side wiggles" of light. If you design your chips correctly, you can use these wiggles to make light-mixing devices much more powerful.
2. It's Simple and Scalable: You can stick these 2D stickers onto standard silicon chips (the kind used in your phone or computer) and get amazing results without complex manufacturing.

In summary: The researchers took a material that was bad at mixing light (silicon), stuck a super-thin, magic sticker on it (MoS₂), and realized that the light interacts with the sticker in a more complex way than anyone thought. By tuning the track the light travels on, they turned a weak signal into a powerful one, opening the door for faster, more efficient optical computers and quantum devices.

Technical Summary

Here is a detailed technical summary of the paper "Vectorial engineering of second-harmonic generation in silicon-based waveguides integrated with 2D materials."

1. Problem Statement

Integrated photonic devices based on silicon (Si) and silicon nitride (SiN) are highly attractive due to CMOS compatibility but inherently lack bulk second-order nonlinear susceptibility ($\chi^{(2)}$) because of their centrosymmetric crystal structures. While strategies like strain engineering or hybrid integration with Lithium Niobate exist, they often add manufacturing complexity or limit scalability.

Transition Metal Dichalcogenides (TMDs), such as monolayer MoS$_2$, offer a solution due to their broken inversion symmetry and large effective $\chi^{(2)}$. However, previous research and modeling of TMD-integrated waveguides have relied on scalar models. These models assume that nonlinear interactions only occur when the dominant electric field component of the optical mode is parallel to the 2D material plane. Consequently, scalar models predict negligible Second Harmonic Generation (SHG) for modes where the main electric field is orthogonal to the 2D material (e.g., Transverse Magnetic or TM modes), failing to account for the full vectorial nature of the electromagnetic fields and the tensorial nature of the TMD's susceptibility.

2. Methodology

The authors developed a comprehensive full vectorial-tensorial model to describe SHG in SiN waveguides integrated with monolayer MoS$_2$.

• Theoretical Framework:

  • They formulated nonlinear coupled equations that account for the full vectorial overlap between pump and signal modes within the MoS$_2$ domain.
  • The model incorporates the $\chi^{(2)}$ tensor of MoS$_2$ (D$_{3h}$ symmetry), which couples field components in the plane ($x, z$) but not the out-of-plane direction ($y$).
  • Crucially, the model includes the longitudinal electric field component ($E_z$) along the waveguide axis. Even for TM modes (where $E_y$ is dominant), the $E_z$ component interacts with the MoS$_2$ lattice, enabling SHG that scalar models miss.
  • The conversion efficiency ($\eta$) was calculated based on the phase mismatch ($\Delta\beta$), interaction length ($L$), and the nonlinear coefficient ($\gamma$), which depends on the angle ($\theta$) between the waveguide axis and the MoS$_2$ armchair direction.

• Experimental Setup:

  • Device Fabrication: SiN waveguides (1 $\mu$m $\times$ 0.8 $\mu$m cross-section) were fabricated by Ligentec. Monolayer MoS$_2$ flakes were exfoliated and dry-transferred onto the waveguides using a PDMS stamp.
  • Characterization: Raman spectroscopy confirmed the monolayer nature of the flakes. Polarization-resolved SHG was used to determine the crystal orientation relative to the waveguide.
  • Measurement: The system was pumped with femtosecond pulses at 1560 nm (fundamental) to generate SHG at 780 nm. Both horizontal (H/TE) and vertical (V/TM) polarizations were tested for pump and signal.

• Phase Matching Design:

  • Using the vectorial model, the authors designed a waveguide with a width of 1.22 $\mu$m to achieve phase matching between the TE$_0$ pump mode and the TM$_2$ signal mode.
  • A second set of experiments utilized a longer interaction length (~110 $\mu$m) with this phase-matched geometry.

3. Key Contributions

1. Vectorial-Tensorial Modeling: The paper establishes that scalar models are insufficient for 2D-material-integrated photonics. It demonstrates that cross-polarized (TE $\to$ TM, TM $\to$ TE) and co-polarized interactions involving TM modes are significant because the longitudinal field component ($E_z$) drives the nonlinear interaction, even when the dominant field is orthogonal to the 2D material.
2. Discovery of TM Mode Dominance: Contrary to scalar predictions, the study reveals that TM signal modes (specifically TM$_2$) provide the highest conversion efficiency because their $E_z$ component overlaps strongly with both TE and TM pump modes within the MoS$_2$ domain.
3. Generalized Design Guidelines: The authors provide a material-agnostic framework applicable to various 2D materials (e.g., WS$_2$, WSe$_2$, ReS$_2$) for optimizing nonlinear frequency conversion in hybrid photonic platforms.

4. Key Results

• Experimental Enhancement:
  • In a standard 1 $\mu$m-wide waveguide (non-phase-matched), integrating MoS$_2$ resulted in SHG enhancements of 2.2$\times$ to 3.1$\times$ compared to bare waveguides, depending on polarization configurations.
  • Notably, SHG was observed and enhanced even when the pump was TM-polarized (orthogonal to the MoS$_2$ plane), validating the vectorial model.
• Phase-Matched Performance:
  • By engineering a 1.22 $\mu$m-wide waveguide for TE$_0$ $\to$ TM$_2$ phase matching, the authors achieved a 14$\times$ enhancement relative to the non-phase-matched case.
  • Compared to free-space (normal incidence) excitation of a MoS$_2$ monolayer on a glass slide, the phase-matched waveguide achieved a 220$\times$ enhancement in conversion efficiency, despite the MoS$_2$ covering only ~4% of the total waveguide length (110 $\mu$m interaction length).
• Angular Dependence: The conversion efficiency showed a complex dependence on the angle $\theta$ between the waveguide and the MoS$_2$ armchair direction, deviating from simple $\cos(3\theta)$ or $\sin(3\theta)$ behaviors due to the superposition of multiple field components in the vectorial overlap integral.

5. Significance

This work fundamentally shifts the understanding of nonlinear optics in 2D-material-integrated photonics. By proving that vectorial effects (specifically longitudinal field components) are critical for efficient SHG, the paper:

• Corrects Design Flaws: Prevents the dismissal of potentially efficient TM-mode interactions in future device designs.
• Enables High Efficiency: Demonstrates that careful modal phase matching and vectorial engineering can overcome the atomic thickness limitations of 2D materials, achieving massive efficiency gains (220$\times$) over free-space excitation.
• Broad Applicability: The proposed framework is not limited to MoS$_2$ or SiN; it provides a universal tool for designing nonlinear photonic devices using any 2D material, paving the way for compact, efficient frequency converters, entangled photon sources, and parametric amplifiers on silicon chips.