Characterization, 1064 nm photon signals and background events of a tungsten TES detector for the ALPS experiment

Abstract

The high efficiency, low background, and single-photon detection with transition-edge sensors (TES) is making this type of detector attractive in widely different types of applications. In this paper, we present first characterizations of a TES to be used in the Any Light Particle Search (ALPS) experiment searching for new fundamental ultra-light particles. Firstly, we describe the setup and the main components of the ALPS TES detector (TES, millikelvin-cryostat and SQUID readout) and their performances. Secondly, we explain a dedicated analysis method for single-photon spectroscopy and rejection of non-photon background. Finally, we report on results from extensive background measurements. Considering an event selection, optimized for a wavelength of 1064 nm, we achieved a background suppression of $\sim {10}^{-3}$ with a $\sim 50$% efficiency for photons passing the selection. The resulting overall efficiency was 23% with a dark count rate of $8.6\times {10}^{-3}{\mathrm{s}}^{-1}$. We observed that pile-up events of thermal photons are the main background components.

Keywords:

• transition-edge sensor
• single-photon detector
• infrared
• background
• dark count rate

View correction statement:

Erratum

Acknowledgements

We want to thank NIST, Boulder, US, for the TES devices and PTB, Berlin, Germany, for the SQUID sensors. We also want to thank our ALPS collaborators, especially S. Ghazaryan, E. von Seggern and C. Weinsheimer.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 National Institute of Standards and Technology.

2 The coating reduces the standard 4%-loss at a fiber end to $<1$%.

3 In the current setup, we operate two TES chips from NIST: channel A and B. In this paper, we only refer to channel A which is characterized in more detail. For channel B, we have found comparable results.

4 Superconducting Quantum Interference Device.

5 Physikalisch-Technische Bundesanstalt.

6 Tektronix DPO7104C${}^{\circledR }$.

7 Nominal wavelength of 405, 532, 635 and 1064 nm.

8 The averaged pulse results from measurements including only events which are lying in the 3$\mathit{\sigma }$-region of the extended two-dimensional PHD. Using the average pulse as the expected photon pulse shape, the limited readout bandwidth of the readout system is taken into account automatically. Furthermore, the expected pulse is a self-consistent estimate of the overall system and does not depend on the first-order TES theory [14].

9 The nominal wavelength of the used laser is 1063.9 nm. The corresponding energy deviation is much smaller than the energy resolution of the TES detector, $\mathrm{\Delta }E/E<10$%.

10 Instead of the ${\mathit{\chi }}^2$, we consider the reduced chi-squared which is defined by ${\mathit{\chi }}_{\mathrm{red}.}^2=\frac{{\mathit{\chi }}^2}{\mathit{\nu }}$, where $\mathit{\nu }$ is the degree of freedom. In this example, it is $\mathit{\nu }=N-n=400-2=398$, which includes all $N=400$ data points of an event window and $n=2$ due to two free fit parameters, the scaling factor $a$ and the shift index $j$.

11 The optical losses are mainly caused by the used fiber setup. Using a commercial power meter, we measured a transmission of only $\sim 60$% due to two fibers with a fiber connector in between. In addition, we conservatively estimate a 95%-efficiency of the fiber-to-TES coupling (Section 2.1). Because of the chosen TL, only $\sim 80$% of the nominal signals are considered. Furthermore, the two-dimensional Gaussian $3\mathit{\sigma }$-region contains 98.9% nominally. Finally, the half $3\mathit{\sigma }$-region includes a factor 0.5.