Highly sensitive detection for infrared photons by non-degenerate two-photon absorption under mid-infrared pumping

This paper demonstrates a highly sensitive, room-temperature infrared photon counting method using non-degenerate two-photon absorption in a silicon avalanche photodiode pumped by a mid-infrared field, which achieves a signal enhancement factor of approximately $10^5$ and significantly reduces noise compared to conventional near-infrared pumping schemes.

The Gist

Imagine you are trying to hear a tiny, whispering bird (an infrared photon) in a room that is already very noisy. Usually, the bird's whisper is so faint that the background noise of the room drowns it out completely. This is the main problem scientists face when trying to detect infrared light, especially at room temperature: the "noise" is too loud, and the "whisper" is too quiet.

This paper describes a clever new trick to solve this problem using a silicon detector (a common type of light sensor) and a special "loudspeaker" technique.

The Problem: The "Whisper" vs. The "Roar"

Infrared light has low energy. To make a silicon detector "see" it, you usually need to hit it with a lot of energy at once.

• The Old Way: Scientists tried to use a strong laser beam (the pump) to boost the signal. However, this strong laser was so powerful that it created its own noise (like a roar in the room), which often drowned out the very signal they were trying to detect. It was like trying to hear a whisper while someone was shouting right next to you.
• The Limitation: If they turned the laser down to reduce the noise, the signal became too weak to detect. If they turned it up, the noise became too loud.

The Solution: The "Two-Person Team"

The researchers developed a new method called Non-Degenerate Two-Photon Absorption (ND-2PA). Here is how it works using a simple analogy:

Imagine the silicon detector is a heavy door that is stuck. You need two people to push it open at the exact same time to make it move.

1. Person A (The Signal): This is the faint infrared light you want to detect. It's too weak to push the door open on its own.
2. Person B (The Pump): This is a strong laser beam.

The Old Strategy (Degenerate): Both people were wearing the same heavy boots (same wavelength). When they pushed, the heavy boots of the "Pump" person created a lot of friction and noise (background noise) that made it hard to tell if the door moved because of the "Signal" person or just the "Pump" person's own movement.

The New Strategy (Non-Degenerate):
The researchers changed the "Pump" person's shoes. They gave them a special pair of shoes (a Mid-Infrared laser at 3 micrometers) that are so light they don't create any friction or noise on their own.

• The "Pump" person can now push as hard as they want without making any noise.
• When the "Signal" person (the faint infrared light) joins in, the door opens.
• Because the "Pump" person is silent, the door only opens when the "Signal" person is there.

The Results: Hearing the Unhearable

By using this "silent pump" strategy, the team achieved some incredible results:

• Silence in the Room: They eliminated the background noise that usually comes from the pump laser. The noise level dropped by a factor of 100 (two orders of magnitude) compared to older methods.
• Super Sensitivity: They could detect infrared pulses so weak they contained only a few thousand photons (energy at the "femtojoule" level).
• The Boost: The ability to count these signals improved by a factor of 100,000 (10 to the power of 5) compared to the old way of doing it.

A Catch: The "Hot Door"

The researchers also noticed something interesting. If the "Pump" person pushes too hard for too long, the door (the detector) gets hot. When it gets hot, it becomes a bit sluggish, and the counting rate starts to drop. This tells them that for the best results, they need to manage the heat, but the method still works incredibly well.

Why This Matters

This new setup is like upgrading from a noisy, old radio to a high-fidelity, silent microphone. It allows for:

• Room Temperature Operation: You don't need to freeze the equipment to make it work.
• No Complex Alignment: It doesn't require the tricky "phase-matching" adjustments that other methods need.
• Broad Vision: It can see a wide range of infrared colors.

The paper suggests this could be used for things like measuring distances from far away (remote ranging), sensing tiny amounts of chemicals (sensitive sensing), looking at biological samples (biochemical imaging), and analyzing trace amounts of substances (trace spectroscopy).

In short, they found a way to make a standard silicon detector "hear" the faintest whispers of infrared light by using a special, silent helper laser that doesn't create any noise of its own.

Technical Summary

Technical Summary: Highly Sensitive Detection for Infrared Photons by Non-Degenerate Two-Photon Absorption Under Mid-Infrared Pumping

Problem Statement
Sensitive infrared detection is critical for applications ranging from laser ranging to biochemical sensing. However, the sensitivity of infrared detectors operating at room temperature is typically limited by large dark noise, a consequence of the narrow semiconductor band gaps required for direct infrared absorption. While cryogenic operation can suppress thermal noise, achieving single-photon sensitivity remains challenging. Conversely, silicon-based avalanche photodiodes (Si-APDs) offer low noise and high efficiency but possess a wide band gap (~1.12 eV) that prevents direct detection of infrared photons.

Existing solutions often employ frequency upconversion techniques. While nonlinear methods like sum-frequency generation offer high sensitivity, they require strict phase-matching conditions, limiting their operational bandwidth. Alternatively, detection schemes based on the material nonlinearity of the detector itself, such as two-photon absorption (2PA), avoid phase-matching requirements and alignment complexities. However, previous demonstrations of 2PA-based infrared detection using degenerate schemes (where signal and pump photons have similar energies) suffered from severe background noise. This noise arises from the degenerate two-photon absorption (D-2PA) of the intense pump field itself, which can saturate the detector and limit the dynamic range.

Methodology
The authors propose and demonstrate a detection scheme utilizing non-degenerate two-photon absorption (ND-2PA) in a Si-APD. The core strategy involves using a strong mid-infrared (MIR) pump field to assist the absorption of a weaker infrared signal photon.

• Experimental Setup: The system utilizes a passively synchronized fiber laser setup generating picosecond pulses at 1030 nm (Yb-doped) and 1550 nm (Er-doped). The MIR pump source at 3070 nm is generated via difference-frequency generation (DFG) in a periodically-poled lithium niobate (PPLN) crystal using the dual-color beams. The 1550 nm beam serves as the signal.
• Physical Mechanism: The pump wavelength (3070 nm, 0.4 eV) is chosen specifically to be lower than half the silicon band gap (0.56 eV). This ensures that the pump photons alone cannot induce D-2PA, effectively eliminating the dominant pump-induced background noise. The combined energy of one signal photon (0.8 eV) and one pump photon (0.4 eV) exceeds the silicon band gap (1.12 eV), enabling electron-hole pair generation via ND-2PA.
• Noise Suppression: By operating in this non-degenerate regime, the background noise is reduced to the level of the much weaker three-photon absorption (3PA) of the pump, rather than the strong D-2PA of the pump.

Key Contributions and Results

1. Elimination of Pump-Induced Noise: The study successfully demonstrated that choosing a pump photon energy ($\hbar\omega_2$) such that $\hbar\omega_2 < E_g/2$ excludes D-2PA for the pump. Experimental results confirmed that the pump-induced noise in this MIR regime is approximately six orders of magnitude lower than in conventional near-infrared (NIR) pumping schemes where D-2PA occurs.
2. Enhanced Sensitivity and Counting Rate: The suppression of background noise allowed for the identification of infrared pulses with femtojoule-level energy (approximately $8 \times 10^3$ photons per pulse). The signal counting rate exhibited an unprecedented enhancement factor of approximately $10^5$ compared to direct detection based on D-2PA.
3. Noise Equivalent Power (NEP) Improvement: The Noise Equivalent Power (NEP) was improved by two orders of magnitude compared to conventional schemes using NIR pumping. Theoretical modeling of the NEP as a function of pump power confirmed that the MIR pumping strategy maintains a significantly lower noise floor.
4. Thermal Effects: The authors investigated the thermal effects of high-intensity MIR pumping. They observed that at high pump powers, the count rate saturates and eventually decreases. This behavior was attributed to thermal effects altering the reverse-biased voltage or absorption coefficients, providing practical guidance for optimizing pump power in future applications.
5. Broadband Potential: Theoretical analysis suggests the scheme is applicable for broadband detection. For instance, with a 3 µm pump, signals up to 1.7 µm can be detected; with a 2.4 µm pump, the range could cover 1.2–2 µm.

Significance and Claims
The paper claims that this configuration offers a viable alternative for sensitive infrared detection and imaging with several distinct advantages:

• Room-Temperature Operation: It leverages the low-noise characteristics of Si-APDs without requiring cryogenic cooling.
• No Phase-Matching Requirement: Unlike upconversion methods, the detection relies on intrinsic material nonlinearity, removing the need for complex optical alignment or narrow acceptance windows.
• Broadband Response: The method is not restricted to specific wavelengths, provided the sum of photon energies exceeds the band gap.

The authors conclude that this approach enables the detection of femtojoule-level pulses and could facilitate applications in remote ranging, sensitive sensing, biochemical imaging, and trace spectroscopy. They further note that while the current sensitivity is high, using even lower pump photon energies (wavelengths > 3300 nm) could theoretically eliminate 3PA noise entirely, potentially leading to single-photon sensitivity. Additionally, the pulsed nature of the pump provides high-precision temporal resolution suitable for time-resolved spectroscopy.