Photodiode quantum efficiency for 2-μm light in the signal band of gravitational wave detectors

TL;DR

The paper addresses the challenge of achieving near-unity true quantum efficiency for photodiodes at 2 µm, a wavelength regime of interest for future squeezed-light gravitational-wave detectors. Using a commercial extended-InGaAs photodiode and 2128 nm light, the authors measure dark noise and detection efficiency across temperatures from room temperature to 4 K, highlighting the distinction between true quantum efficiency and conventional detection efficiency. They find that cooling reduces dark noise substantially (by more than 15 dB around 180 K) but also reduces the photodiode's detection efficiency, with about a 15% drop at 250 K and severe degradation at very low temperatures due to band-gap effects. The study concludes that cooling alone cannot deliver the required high QE and calls for a new photodiode design with QE > 99% at low frequencies for 2 µm light, guiding future development for cryogenic GW detectors.

Abstract

Quantum technologies with quantum correlated light require photodiodes with near-perfect `true' quantum efficiency, the definition of which adequately accounts for the photodiode dark noise. Future squeezed-light-enhanced gravitational wave detectors could in principle achieve higher sensitivities with a longer laser wavelength around 2 μm. Photodiodes made of extended InGaAs are available for this range, but the true quantum efficiency at room temperature and the low frequency band of gravitational waves is strongly reduced by dark noise. Here we characterize the change in performance of a commercial extended-InGaAs photodiode versus temperature. While the dark noise decreases as expected with decreasing temperature, the detection efficiency unfortunately also decreases monotonically. Our results indicate the need for a dedicated new design of photodiodes for gravitational wave detectors using 2-μm laser light.

Figures (3)

• Figure 1: A Thorlabs FD05D photodiode (PD) was placed in a helium cryostat and cooled down to 4 K within 48 h. The photo current was converted to a voltage by a high-gain self-built transimpedance amplifier located in room temperature environment. During cool-down and warm-up, the laser light on the photodiode was switched on and off every 5 minutes and the photo voltage values were continuously sampled and stored at a rate of 1 kHz. This measurement data provided dark current, total photo current and their spectral densities. In parallel, the temperature of the photodiode was measured with a sensor in a common copper block. The stability of the light power on the photodiode was ensured by automatic beam centering and checked by a separate measurement on the input beam. DAQ: data acquisition system.
• Figure 2: Noise power spectral densities from 1 to 100 Hz. Left: Comparison of laser power noise from 0.1 mW at 2128 nm and dark noise measured with the photodiode FD05D at 144 K for three bias voltages (The resolution bandwidth, RBW: 0.1 Hz, no averaging) Right: Photodiode dark-noise at four temperatures, normalized to that at room temperature. With decreasing temperature, detector dark noise strongly decreased as expected. The photodiode bias voltage was 0.39 V. (RBW: 0.25 Hz, 50 times averaged).
• Figure 3: Change of the detection efficiency of PD FD05D with temperature, normalized to that at room temperature. The data was collected continuously over one full cooling cycle. Blue points were taken while cooling down over a time frame of 48 hours. Red points show the natural warming up over 16 days (cooling switched off, no heating). The missing data in the warm-up is due to a computer failure. A small hysteresis of the data was observed. The shaded areas indicate our estimation of systematic error bars due to fluctuations in the light power. All data was measured at a photodiode bias voltage of 0.39 V.