Objective Compared with single-photon detectors (SPDs) based on superconducting nanowires or photon up-conversion, SPDs based on InGaAs(P)/InP single-photon avalanche diodes (SPADs) have shown advantages such as small size, low power consumption, and low cost. Therefore, they have been widely used in the fields of lidar, three-dimensional imaging, quantum information, etc. With the development of applications, the overall performance of InP-based SPDs has been gradually improved these years. However, striking a balance among detection efficiency, dark count rate, afterpulse probability, and dead time is still challenging. High afterpulse probability is found to be a bottleneck of the performance for InP-based free-running SPDs, and the dead time has to be set to a large value to suppress the severe after-pulsing effects. Besides, the requirement of compact SPDs for the use in unmanned platforms, vehicles and integrated systems is increasing. However, the integration of SPADs is often accompanied with a degradation in some of the performance specifications or parameter adjustment flexibility. In this contribution, an integrated SPD based on InGaAsP/InP SPAD with fast active quenching was developed for 1.06 μm, size-limited, and low-dead-time applications.
Methods A fast active-quenching circuit was proposed to cease the avalanche current of the SPAD quickly and actively, in order to reduce the number of avalanche carriers, and consequently the afterpulse probability. The circuit is essentially composed of only an ultra-high-speed Si-Ge comparator and a GaAs enhancement-mode pseudo-morphic high electron mobility transistor within the feedback loop. The latch-enable function of the high-speed comparator acts as the hold-off logic, and hence the delay of the feed-back loop is minimized. An improved C-RC network is used to cancel the noise introduced by the quenching signal. By integrating the critical balancing capacitor in the C-RC network into the package of the SPAD, the discrimination threshold of the comparator can be set as low as 2.4 mV. With all the efforts above, the quenching delay is minimized, and hence the full width at half-maximum of the avalanche current was only approximately 250 ps. In addition, the detector has integrated a negative-high voltage generation circuit, a thermo-electric cooler control circuit, an FPGA-based logic control circuit. All the printed circuit boards of the above circuits were smaller than 33 mm×40 mm, and were stacked to achieve a small size. With the optimization of the quenching circuits and the integration of the thermal-electric cooler, the detector has achieved high performance, compact size, and low power consumption at the same time. It has a compact size of only 63 mm×54 mm×44 mm, with 105 μm /125 μm pig-tailed multi-mode fiber for easy coupling, and has embedded gating, parameter control, and time-correlated single photon counting (TCSPC) features. Besides, a TCSPC system was built for performance evaluation of the proposed detector, including dark count rate, detection efficiency, total afterpulse probability, and time jitter.
Results and Discussions The proposed detector has good overall performance. The single-photon detection efficiency reached 30%, and the time jitter (FWHM) was 329, 200, 162, 146, and 139 ps at the detection efficiencies of 10%, 15%, 20%, 25%, and 30%, respectively. As the discrimination threshold was close to the limit of the comparator, a lower time jitter can only be achieved by increasing the excess bias. The dark-count rate was generally low at -30 ℃: the values at the detection efficiencies of 10%, 20%, and 30% were 0.9, 2.7, 7.5 kHz, respectively. The dark count rate was approximately doubled with every 10 ℃ increment of temperature. With the increase of the photon detection efficiency, the dark count rate rose exponentially, mainly contributed by trap-assisted tunneling at the InP multiplication layer in the SPAD. The dark counts increased even more quickly at lower dead time due to higher afterpulse probability. Most importantly, the minimum dead time of the detector was as low as 50 ns. At the temperature of -30 ℃, the detection efficiency of 10%, and the dead time of 50 ns and 100 ns, the total afterpulse probability was measured to be approximately 15% and 10%, respectively. Such low dead time and low afterpulse probability at the detection efficiency of 10% enable its future use in practical lidar systems. As for the condition of the detection efficiency higher than 20%, the dead time was set above 1 μs to achieve sufficiently low afterpulse probability. For example, the afterpulse probability was 15%--20% at the detection efficiency of 30%, dead time of 2 μs. Higher cooling temperature could reduce the afterpulse probability at the cost of higher dark count rate. Besides, the detectors have demonstrated low power consumption. The total power consumption was 4.0, 4.8,and 5.4 W when cooling to -10, -20,-30 ℃, respectively, where 2.6 W was contributed by the circuits excluding the cooling part.
Conclusions In this paper, an integrated InGaAsP SPD for 1.06 μm was presented. With the optimization of the quenching circuits and the integration of the thermal-electric cooler and key components for quenching, the detector has achieved ultra-low quenching delay, compact size, and low power consumption at the same time. The minimum dead time was as low as 50 ns, where the dark-count rate and afterpulse probability were approximately 1 kHz and 15%, respectively, at the photon detection efficiency of 10% and the temperature of -30 ℃. The detector has 105 μm /125 μm multi-mode fiber coupling and a compact size of only 63 mm×54 mm×44 mm. The low dead time, small size and easy-to-use features are making the detector especially suitable for use in size-limited single-spot and multi-beam lidar.