🔬 optics

Long-wave infrared Fourier transform spectroscopy with enhanced and scalable sensitivity

This paper presents a broadband long-wave infrared Fourier transform spectrometer that integrates dual-comb spectroscopy, electro-optic sampling, and multi-channel near-infrared detection to achieve unprecedented sensitivity (0.3 ppb for NH $_3$ and 2 ppb for C $_2$ H $_4$ ) and high-resolution, scalable analysis of complex gas mixtures.

Original authors: Sergey Vasilyev, Roderik Krebbers, Dmitrii Konnov, Mathieu Walsh, Igor Moskalev, Mike Mirov, Andrey Muraviev, Jerome Genest, Konstantin Vodopyanov, Simona M. Cristescu

Published 2026-04-14

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Sergey Vasilyev, Roderik Krebbers, Dmitrii Konnov, Mathieu Walsh, Igor Moskalev, Mike Mirov, Andrey Muraviev, Jerome Genest, Konstantin Vodopyanov, Simona M. Cristescu

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 listen to a very quiet conversation happening in a noisy, crowded room. You want to hear exactly what everyone is saying, even the whispers, without missing a single word. This is essentially what scientists do when they try to detect tiny amounts of gas in the air.

This paper describes a new, super-powerful "listening device" for gases that is far more sensitive than anything that existed before. Here is how they did it, explained in simple terms:

1. The Problem: The "Faint Whisper" in the Dark

Most gases leave a unique "fingerprint" in the Long-Wave Infrared (LWIR) part of the light spectrum. Think of this as a secret code that only specific gases can read. To find a gas, you shine light through the air and look for which parts of the light get "eaten" (absorbed) by the gas molecules.

The problem is that LWIR light is hard to catch. The detectors (cameras) we usually use for this are like old, grainy film cameras. They are expensive, need to be frozen with liquid nitrogen to work, and they are often too noisy to hear the faintest whispers (trace gases).

2. The Solution: The "Translator" Trick

The team built a machine that acts like a universal translator. Instead of trying to catch the difficult LWIR light directly, they do something clever:

They generate a special laser beam (the "source") that is incredibly quiet and powerful.
They mix this beam with the gas sample.
Then, they use a technique called Electro-Optic Sampling (EOS). Imagine this as a magic bridge that instantly translates the invisible, hard-to-catch LWIR light into Near-Infrared (NIR) light.

Why is this a big deal? Because NIR light is easy to catch! We have cheap, high-quality, room-temperature detectors (InGaAs photodiodes) for NIR light that are like high-definition digital cameras. They are 100 to 1,000 times better at seeing faint signals than the old LWIR detectors.

The Analogy: It's like trying to hear a whisper in a storm. Instead of straining your ears (using a bad detector), you use a microphone that converts the sound into a clear text message on your phone (the NIR detector) that you can read perfectly.

3. The "Super-Resolution" Camera

This new machine doesn't just hear better; it sees in super-high definition.

Old machines might see a blurry blob where a gas is.
This new machine has a resolution so sharp it can distinguish between two gas molecules that are standing right next to each other. It's like going from a standard definition TV to an 8K Ultra HD screen.

4. The "Multi-Channel" Upgrade: One Ear vs. Four Ears

Usually, these machines use one detector. But the authors realized that since the "translated" light is so bright and easy to handle, they could split it up!

They used four detectors working at the same time.
Analogy: Imagine trying to hear a faint sound. If you have one ear, you might miss it. If you have four ears listening in perfect sync, you can hear it much better.
By splitting the signal among four detectors, they doubled their sensitivity again, making the machine even more powerful without making it much more complicated.

5. The Results: Hearing the Unhearable

With this new setup, they achieved amazing results:

Ammonia (NH3): They could detect just 0.3 parts per billion. That is like finding a single grain of sand in a large swimming pool.
Ethylene (C2H4): They could detect 2 parts per billion.
They did this while looking at a wider range of colors (gases) and with sharper detail than any previous machine.

6. The Real-World Test: "Breath" Analysis

To prove it works, they tested a mixture of gases that mimics human breath.

They successfully found methane, methanol, isoprene, and even tiny traces of ammonia and water vapor.
They could tell exactly how much of each gas was there, even though some were hiding in the "noise" of the others.
This proves the machine could one day be used to diagnose diseases by analyzing a patient's breath, or to detect dangerous leaks in factories instantly.

Summary

In short, this paper is about building a super-sensitive gas detector by:

Using a super-stable laser.
Translating hard-to-detect infrared light into easy-to-detect light.
Using cheap, high-quality cameras to catch that light.
Using four cameras at once to boost the signal.

The result is a machine that can "see" invisible gases with a clarity and sensitivity that was previously thought impossible, opening the door to better medical diagnostics, environmental monitoring, and industrial safety.

1. Problem Statement

Fourier Transform Spectroscopy (FTS) and Dual-Comb Spectroscopy (DCS) are powerful tools for analyzing the Long-Wave Infrared (LWIR) "molecular fingerprint" region (5–20 µm). However, their sensitivity in practical applications has historically been limited by several factors:

Source Limitations: Achieving high sensitivity requires light sources with simultaneously high power and ultra-low intensity noise, a combination difficult to realize in the LWIR.
Detector Limitations: Traditional LWIR detectors (e.g., HgCdTe, DLaTGS) often suffer from lower detectivity ( $D^*$ ) and require cryogenic cooling, limiting the dynamic range and signal-to-noise ratio (SNR).
Scalability: While parallel detection using multiple detectors could theoretically improve sensitivity, practical implementations are limited by system complexity and the difficulty of splitting broadband LWIR signals. Previous attempts were restricted to $N_D = 2$ detectors.
Performance Gap: Existing high-sensitivity LWIR instruments often sacrifice spectral resolution or bandwidth. Conversely, high-resolution instruments often lack the sensitivity required for trace gas detection (parts-per-billion, ppb).

2. Methodology

The authors developed a novel Electro-Optic Sampling Dual-Comb Spectrometer (EOS-DCS) that bridges LWIR spectroscopy with Near-Infrared (NIR) photonics to overcome these limitations.

Ultra-Low Noise Source: The system utilizes a Cr:ZnS (Chromium-doped Zinc Sulfide) optical frequency comb.
- Technology: Based on a polycrystalline Cr:ZnS master oscillator and power amplifier.
- Noise Suppression: The system leverages Random Quasi-Phase-Matching (RQPM) within the polycrystalline gain medium. This nonlinear process suppresses intensity noise spikes and provides self-defocusing/self-phase modulation, resulting in an integrated RMS intensity noise of <0.1% and a linewidth of 7.2 kHz.
- Stabilization: The combs are phase-locked to an atomic clock (Rb) and an optical reference, ensuring fractional stability $<7 \times 10^{-15}$ .
Frequency Conversion:
- LWIR Generation: The fundamental 2.4 µm comb is down-converted to the LWIR (6.5–12 µm) via Optical Rectification (OR) in a ZGP crystal (~0.3 W power).
- NIR Generation: A portion of the comb is up-converted to NIR (1.2 µm) via Second Harmonic Generation (SHG) in a PPLT crystal (~1 W power).
Electro-Optic Sampling (EOS):
- The LWIR comb (sampled through a 31.2 m multi-pass gas cell) and the NIR comb are mixed in an EOS module.
- This process maps the electric field of the LWIR signal onto the NIR domain via sum-frequency generation.
- Key Advantage: This allows the use of high-performance, room-temperature InGaAs photodiodes (standard in telecom) instead of expensive, cryogenically cooled LWIR detectors. InGaAs detectors offer $10^1$ – $10^2$ times higher detectivity than HgCdTe and $10^4$ times higher than DLaTGS.
Parallel Acquisition & Processing:
- The high-power NIR signal is split and detected by four balanced InGaAs photodiodes simultaneously.
- A real-time GPU-based processing unit (PUG) performs computational corrections, including phase noise retrieval, timing jitter correction, interferogram alignment, and averaging.

3. Key Contributions

Architectural Innovation: Successfully demonstrated the first EOS-DCS system capable of scalable sensitivity via parallel detection in the NIR domain, overcoming the complexity barriers of parallel LWIR detection.
Source Advancement: Validated the Cr:ZnS laser as a superior source for high-performance spectroscopy due to its ultra-low intensity noise (enabled by RQPM) and high peak power, facilitating efficient nonlinear conversion.
Sensitivity Scaling: Proved that sensitivity scales with the square root of the number of detectors ( $\sqrt{N_D}$ ) in a symmetric division regime, achieving a 40x improvement over previous LWIR demonstrations.
Computational Correction: Implemented real-time GPU-based correction of interferograms, effectively removing phase noise and drift, enabling long-term coherent averaging.

4. Key Results

Detection Limits (in 500 seconds):
- Ammonia (NH $_3$ ): 0.3 ppb (improvement of 20x over previous LWIR records).
- Ethylene (C $_2$ H $_4$ ): 2 ppb (improvement of 40x over previous LWIR records).
Performance Metrics:
- Spectral Resolution: 0.0027 cm $^{-1}$ (significantly higher than the 0.1 cm $^{-1}$ of previous high-sensitivity LWIR FTS).
- Spectral Window: 800 – 1400 cm $^{-1}$ (2x broader than previous records).
- Figure of Merit (FOM): The EOS-DCS system achieved an FOM 160 times higher than a state-of-the-art FTS instrument when using four detectors.
- Stability: The system maintained linear $\Sigma$ SNR scaling with $\sqrt{t}$ for up to 20,000 seconds, whereas the reference FTS deviated after ~500 seconds.
Complex Mixture Analysis:
- The system successfully analyzed a synthetic "human breath" mixture containing CH $_4$ , CH $_3$ OH, C $_5$ H $_8$ , (CH $_3$ ) $_2$ CO, CO $_2$ , and trace NH $_3$ .
- It quantified concentrations of 0.7 ppm isoprene, 0.7 ppm methanol, 65 ppb ammonia, and 2 ppm methane within 103 seconds, demonstrating the ability to resolve trace species in the presence of strong absorbers (CO $_2$ , H $_2$ O).

5. Significance

This work establishes a new performance envelope for LWIR spectroscopy by decoupling the detection mechanism from the spectral range. By mapping LWIR information to the NIR domain, the authors leverage mature, high-performance, and scalable NIR detector technology.

Scalability: The architecture allows for straightforward sensitivity improvements simply by increasing the number of detectors (linear scaling with proper spectral de-multiplexing), paving the way for parts-per-trillion (ppt) detection on minute timescales.
Application Impact: The combination of sub-ppb sensitivity, high spectral resolution, and broad bandwidth makes this system ideal for real-time, multispecies analysis of complex gas mixtures. This is critical for applications in environmental monitoring, industrial safety, medical diagnostics (breath analysis), and fundamental molecular spectroscopy.
Technological Shift: It challenges the notion that high-sensitivity LWIR spectroscopy requires cryogenic cooling or complex cavity enhancement, offering a robust, room-temperature alternative with superior speed and resolution.