Fig. 1. Device layout. (a) Scanning electron microscope image showing the nanowire bends. The image shows 64 electrode lines arranged in a line, with an active area of a 28 μm diameter (green), which is the main photosensitive surface of the device. During optical coupling, the light spot is focused on this area. (b) Enlarged image of the boxed area in (a). Each pixel is parallel to each other. Different nanowires have different colors. (c) Enlarged image of the boxed area in (b). Each nanowire has a line width and period of 70 and 140 nm, respectively.

Fig. 2. Device testing layout. (a) Light emitted from a tunable pulse laser vertically incident on the device area after passing through two attenuators and a three-ring polarization controller. The device is placed in our self-built 64-channel system. The system’s output is read using four electronic devices capable of biasing 16 channels simultaneously. The pulses from the readout can be directly displayed by the software provided with the device. Laser: in this experiment, three types of laser light sources were used, a femtosecond fiber laser (Calmar, FPL-01CAF) for efficiency testing, a continuous-wave tunable laser (Keysight, 81970A) for counting rate determination, and a tunable pulse laser (Anhui Quantum Communication Co., Ltd.) for PNR capacity testing. AT1/AT2: variable attenuator; SMF: single-mode fiber; 64-channel bias and readout: the device integrates 64-channel bias modules that can achieve simultaneous bias and signal amplification for the 64 channels. The device can amplify the input signal and process it into a square-wave signal output. (b) Photo of the packaged device placed on the second stage of the temperature-controlled platform in our self-built system. Sixty-four homemade flexible coaxial cables were deployed in the cooling system. The leakage heat testing results indicate that all the leakage heat values of the self-made flexible cable were 4 mW.

Fig. 3. Device characterization. (a) Distribution of switching currents for each pixel and normalized count rate to total input for each pixel. The red curve in the figure is a Gaussian fit to the efficiency distribution, showing an approximate Gaussian distribution of the spot. (b) Efficiency test curves of the device: the black dot is the efficiency curve of a single pixel, and the red dot represents the sum of the efficiencies of all pixels. (c) Device CR curves: the red curve represents the CR of a single pixel, and the black curve represents the sum of the CRs of all pixels. In this article, three different types of lines (shortest, medium, and longest) were selected to measure their pulse waveforms, and their recovery times are approximately the same. The inset shows the pulse image of three different device channels. The dead time τd (defined as the time at which the height of the pulse is reduced to 1/e=0.368 of its initial value) is 4.8 ns.

Fig. 4. Photon number resolution characteristics of the device. (a) Measurement of merged output counting statistics at different light intensities. The X coordinate is the pulse amplitude, the Y coordinate is the light intensity, and the Z coordinate is the normalized counting intensity information. (b) Linear relationship between the resolved photon number and the output pulse amplitude. The X axis represents the photon number resolved by the detector, the left Y axis is the center position of the corresponding Gaussian waveform, and the right Y axis is the FWHM of the Gaussian waveform.