Abstract: Traditionally, photon detectors are operated in a direct detection mode counting incident photonswith a known quantum efficiency. This procedure allows one to detect weak sources of radiation but allthe information about its frequency is limited by the optical filtering/resonating structures used which arenot as precise as would be required for some practical applications. In this work we propose heterodynereceiver based on a photon counting mixer which would combine excellent sensitivity of a photon countingdetector and excellent spectral resolution given by the heterodyne technique. At present, Superconducting-Nanowire-Single-Photon-Detectors (SNSPDs) [1] are widely used in a variety of applications providing thebest possible combination of the sensitivity and speed. SNSPDs demonstrate lack of drawbacks like highdark count rate or autopulsing, which are common for traditional semiconductor-based photon detectors,such as avalanche photon diodes.In our study we have investigated SNSPD operated as a photon counting mixer. To fully understandits behavior in such a regime, we have utilized experimental setup based on a couple of distributedfeedback lasers irradiating at 1.5 micrometers, one of which is being the Local Oscillator (LO) and theother mimics the test signal [2]. The SNSPD was operated in the current mode and the bias currentwas slightly below of the critical current. Advantageously, we have found that LO power needed for anoptimal mixing is of the order of hundreds of femtowatts to a few picowatts, which is promising for manypractical applications, such as receiver matrices [3]. With use of the two lasers, one can observe thevoltage pulses produced by the detected photons, and the time distribution of the pulses reproduces thefrequency difference between the lasers, forming power response at the intermediate frequency which canbe captured by either an oscilloscope (an analysis of the pulse statistics is needed) or by an RF spectrumanalyzer. Photon-counting nature of the detector ensures quantum-limited sensitivity with respect to theoptical coupling achieved. In addition to the chip SNSPD with normal incidence coupling, we use thedetectors with a travelling wave geometry design [4]. In this case a NbN nanowire is placed on the topof a Si3N4 nanophotonic waveguide, thus increasing the efficient interaction length. For this reason it ispossible to achieve almost complete absorption of photons and reduce the detector footprint. This reducesthe noise of the device together with the expansion of the bandwidth. Integrated device scheme allowsus to measure the optical losses with high accuracy. Our approach is fully scalable and, along with alarge number of devices integrated on a single chip can be adapted to the mid and far IR ranges wherephoton-counting measurement may be beneficial as well [5].Acknowledgements: This work was supported in part by the Ministry of Education and Science of theRussian Federation, contract No. 14.B25.31.0007 and by RFBR grant No. 16-32-00465.

Abstract: Achievement of the ultimate sensitivity along with a high spectral resolution is one of the frequently addressed problems, as the complication of the applied and fundamental scientific tasks being explored is growing up gradually. In our work, we have investigated performance of a superconducting nanowire photon-counting detector operating in the coherent mode for detection of weak signals at the telecommunication wavelength. Quantum-noise limited sensitivity of the detector was ensured by the nature of the photon-counting detection and restricted by the quantum efficiency of the detector only. Spectral resolution given by the heterodyne technique and was defined by the linewidth and stability of the Local Oscillator (LO). Response bandwidth was found to coincide with the detector’s pulse width, which, in turn, could be controlled by the nanowire length. In addition, the system noise bandwidth was shown to be governed by the electronics/lab equipment, and the detector noise bandwidth is predicted to depend on its jitter. As have been demonstrated, a very small amount of the LO power (of the order of a few picowatts down to hundreds of femtowatts) was required for sufficient detection of the test signal, and eventual optimization could lead to further reduction of the LO power required, which would perfectly suit for the foreseen development of receiver matrices and the need for detection of ultra-low signals at a level of less-than-one-photon per second.