🔬 optics

Ultra-broadband, Low-loss Wavelength Combiners and Filters: Novel Designs and Experiments in Thin-film Lithium Niobate

This paper presents novel closed-form analytical models and experimental demonstrations of compact, ultra-broadband, and low-loss wavelength combiners and filters on a thin-film lithium niobate platform, achieving insertion losses below 0.1 dB and high extinction ratios across both fundamental and second-harmonic wavelengths to enable high-fidelity quantum photonic circuits.

Original authors: Robert Kwolek, Parash Thapalia, Aditya Tripathi, Pooja Kulkarni, Jaber Balalhabashi, Farzaneh Arab Juneghani, Michael Bullock Oanh Hoang Vo, Sasan Fathpour, Rajveer Nehra

Published 2026-03-31

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Robert Kwolek, Parash Thapalia, Aditya Tripathi, Pooja Kulkarni, Jaber Balalhabashi, Farzaneh Arab Juneghani, Michael Bullock Oanh Hoang Vo, Sasan Fathpour, Rajveer Nehra

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 running a busy, high-speed train station, but instead of trains, you are managing beams of light. This is the world of photonic circuits, which are the "chips" that power future computers and quantum computers.

In this paper, the researchers from the University of Massachusetts Amherst and the University of Central Florida have built a new, super-efficient "traffic controller" for these light beams. Here is the breakdown of their invention in simple terms:

The Problem: The "Traffic Jam" of Light

In modern quantum computers, we need to send different colors (wavelengths) of light through a tiny chip at the same time.

The "Fundamental" Light (FH): Think of this as a slow, heavy freight train (red light, 1550 nm) carrying the main data.
The "Second Harmonic" Light (SH): Think of this as a fast, sleek sports car (blue light, 775 nm) used for specific quantum tasks.

Usually, when you try to merge these two different "vehicles" onto the same track or split them apart, you lose a lot of energy. It's like trying to merge two lanes of traffic onto one, but half the cars crash or get stuck. In the world of quantum computing, even a tiny bit of lost energy (loss) destroys the delicate information being carried.

Old methods to do this were like using a long, winding road to merge the lanes gently. It worked, but the road was too long, causing friction (loss) and taking up too much space on the chip.

The Solution: The "Fast-Adiabatic" Merge

The researchers created a new design called a FAQUAD coupler (Fast Quasi-Adiabatic Driving).

The Analogy: The Perfect Dance Move
Imagine two dancers (the light beams) who need to swap partners without ever tripping or losing rhythm.

Old Way: They walk slowly down a long hallway, holding hands, slowly turning until they swap. It's safe, but it takes forever and they get tired (energy loss).
The New FAQUAD Way: The researchers figured out a precise mathematical "dance routine." They tell the dancers exactly how fast to turn and how far to step at every single millisecond. They move fast, but they move so smoothly that they never stumble.

They achieved this by designing the path of the light with three specific sections:

The Straightaway: Where the light beams get close together.
The Cubic Curve: A specially shaped bend (like a smooth S-curve) that guides the beams apart without jerking them.
The Euler Exit: A final gentle curve that sends the beams on their separate ways without any sudden stops.

The Material: The "Super-Highway"

They built this on a material called Thin-Film Lithium Niobate (TFLN).

Analogy: If silicon (used in regular computer chips) is a paved country road, TFLN is a magical, frictionless super-highway. It allows light to move incredibly fast and interact with electricity in ways other materials can't. This makes it perfect for the complex "dance" the researchers designed.

The Results: A Miracle of Efficiency

The team tested their new "traffic controller" and the results were stunning:

Ultra-Low Loss: They lost less than 0.04 decibels of energy.
- Real-world comparison: If you were shouting a message through a 100-foot tube, this device would be so efficient that you'd barely hear a difference in volume at the other end. In fact, it's so efficient that it's almost like the light didn't pass through a device at all.
Broadband: It works perfectly across a wide range of colors (wavelengths), not just one specific shade. It's like a traffic light that works for every car color, not just red ones.
High Precision: It keeps the "freight trains" and "sports cars" completely separate when they need to be, with a separation quality (extinction ratio) of over 25 dB.

Why Does This Matter?

This isn't just about saving a little energy. In Quantum Computing, information is stored in fragile states of light. If you lose even a tiny bit of that light, the quantum information vanishes, and the computer fails.

By creating a device that merges and splits light with almost zero loss, the researchers have built a critical building block for the next generation of quantum computers. They've shown that we can now route complex light signals on a tiny chip without breaking the delicate quantum data, paving the way for powerful, large-scale quantum networks.

In short: They invented a "magic merge lane" for light that is so smooth and efficient, it barely slows down the traffic, allowing the future of quantum computing to move at full speed.

1. Problem Statement

As integrated photonic circuits scale in complexity for quantum and classical applications, the efficient routing and filtering of optical fields across multiple wavelengths (specifically the fundamental harmonic, FH, and second harmonic, SH) become critical.

The Challenge: Nonlinear quantum photonic circuits often require simultaneous handling of pump and signal fields with octave-scale spectral separation (e.g., 1550 nm and 775 nm).
Limitations of Current Tech:
- Silicon Nitride/Silica: While mature, their weak third-order nonlinearity ( $\chi^{(3)}$ ) relies on resonant enhancement, creating a trade-off between interaction strength and bandwidth.
- Conventional Adiabatic Couplers: Standard directional couplers used in Thin-film Lithium Niobate (TFLN) often require long interaction lengths to maintain adiabaticity, leading to high propagation losses.
- Fast Quasi-Adiabatic Driving (FAQUAD): Previous FAQUAD implementations focused only on the constant-gap interaction region, neglecting transition regions. Imperfections in these transitions degrade extinction ratios and increase losses, which is detrimental to quantum coherence.

2. Methodology and Design Framework

The authors propose a novel design for compact, ultra-broadband wavelength combiners and filters operating simultaneously at FH (1550 nm) and SH (775 nm) on a 300-nm TFLN platform.

Analytical Modeling:
- The authors developed closed-form analytical models based on coupled-mode theory.
- They extended the FAQUAD approach to the entire device evolution, not just the central interaction region.
- Key Innovation: To overcome the limitation where driving the mixing angle ( $\chi$ ) to asymptotic values requires impractically large waveguide width differences, they introduced a cubic bend separation profile (Region II) following a constant-gap section (Region I).
- This cubic profile, derived from Euler bend principles, allows the waveguides to separate gradually while maintaining the adiabaticity parameter ( $\eta$ ) constant, ensuring negligible residual coupling at the exit.
Device Geometry:
- Region I: Constant-gap section ( $g_m = 800$ nm) for controlled supermode evolution.
- Region II: Cubic bend section where waveguides separate, preserving adiabaticity.
- Region III: Euler bends for low-loss routing and curvature-optimized transitions.
- Parameters: Waveguide width $\sim 1.2$ µm, sidewall angle $65^\circ$ , etch depth $\sim 100$ nm.
Simulation:
- Full-wave Finite Difference Time Domain (FDTD) simulations were performed using Tidy3D.
- The design was optimized for an adiabaticity parameter $\eta \approx 0.189$ , with a central interaction length $l_m = 264$ µm and a final gap $g_c = 1.2$ µm.

3. Key Contributions

Closed-Form Analytical Solution: The paper provides a mathematical framework linking FAQUAD parameters directly to device geometry and fabrication constraints, allowing for the design of couplers with arbitrary power-splitting ratios.
Full-Evolution Optimization: Unlike previous works, this design optimizes the transition regions (separation bends) to prevent non-adiabatic coupling, significantly improving extinction ratios.
Dual-Wavelength Operation: The device functions as both a wavelength combiner (combining FH and SH) and a filter (separating them) with ultra-low loss at both wavelengths simultaneously.
Experimental Validation: Successful fabrication and characterization on a TFLN platform, demonstrating performance that matches theoretical predictions.

4. Experimental Results

The devices were fabricated on a 300-nm x-cut TFLN wafer using electron-beam lithography and ICP-RIE etching. Performance was characterized using resonator-based transmission measurements.

Fundamental Harmonic (FH, 1550 nm):
- Extinction Ratio: Exceeds 25 dB across a 90 nm bandwidth.
- Insertion Loss: Total added loss is < 0.1 dB across the 1550–1600 nm range.
- Minimum Loss: 0.04 ± 0.02 dB at ~1580 nm.
- Fabrication Tolerance: Robust against etching depth deviations (50 nm) and waveguide width variations (100 nm), with losses remaining < 0.21 dB.
Second Harmonic (SH, 775 nm):
- Extinction Ratio: > 19 dB over a 45 nm bandwidth.
- Insertion Loss: Total added loss is < 0.12 dB over the 45 nm bandwidth.
- Minimum Loss: 0.021 ± 0.003 dB at 775 nm.
Comparison: The results represent an order-of-magnitude improvement in extinction-ratio-limited loss and bandwidth compared to conventional adiabatic designs.

5. Significance

This work addresses a critical bottleneck in scaling quantum photonic circuits: the need for low-loss, broadband routing of multi-wavelength fields.

Quantum Photonic Circuits: By achieving ultra-low loss (< 0.1 dB) and high extinction ratios, these devices preserve quantum coherence and minimize loss-induced degradation in nonlinear processes.
Scalability: The compact footprint and high fabrication tolerance make these components suitable for large-scale, programmable photonic integrated circuits (PICs).
Platform Advancement: The successful demonstration on TFLN highlights the platform's capability to support high-efficiency, single-pass traveling-wave nonlinear devices, bridging the gap between bulk LN devices and integrated circuits.

In conclusion, the authors have demonstrated that FAQUAD-engineered couplers with cubic bends provide a robust, high-performance solution for wavelength routing in TFLN, serving as a foundational building block for next-generation quantum and classical photonic systems.