Single-photon detection at 1550-nm with a high repetition rate was realized using an InGaAs/InP avalanche photodiode operated with a sine wave gating. Removing the AC noise due to the transferred gate signal using band elimination filters, we have discriminated the avalanche signal which is much smaller than that in the conventional gating, which results in the suppression of the afterpulsing. At the repetition frequency of 800MHz, the overall afterpulsing probability was 6.0% with the detection efficiency of 8.5%and the dark count probability of 9.2×10-6.
©2006 Optical Society of America
A single-photon detector (SPD) is an essential component for the optical quantum information processing technologies. In the quantum communication, especially the quantum key distribution (QKD) based on the optical fiber links , it is important to develop the SPD at the telecommunication wavelength (1550-nm) with a high quantum efficiency and low dark counts. Moreover, in practical use, the SPD is required to operate with a high repetition rate. Avalanche photodiodes (APDs) in Geiger mode have been widely used for the SPD. A photo-excited carrier grows into a macroscopic current output via the carrier avalanche multiplication in the APD. Fractions of the many carriers trapped in the APD are subsequently emitted, and trigger additional avalanches. The phenomenon is well known as “afterpulsing” which disturbs single-photon detection. The candidate for the SPD at 1550-nm is an InGaAs/InP-APD. However, the APD in Geiger mode is very noisy, resulting from the afterpulsing and the dark counts caused by the thermal carriers. Therefore, the APD is generally cooled and operated with a gated mode. The gate duration (gate-on time) is generally set to a few nanoseconds in order to reduce the dark count probability per gate. Then the interval between a gate and the subsequent one is set to more than the lifetime of the trapped carriers so that the afterpulsing is suppressed. Using the cooled InGaAs/InP-APD operated with the short gated mode, the quantum efficiency at 1550-nm have reached 10~30% with the dark cont probability per gate of 10-6~10-4 [2, 3, 4, 5, 6]. However, the repetition rate of the gate, which allows the afterpulsing-free single-photon detection, was limited to only ~1MHz because of a long lifetime of the trapped carriers. On the other hand, the SPD using a wavelength up-conversion device and a Si-APD was reported . This scheme doesn’t require the gating operation, and the maximum count rate can reach ~15MHz. However, it is also limited by the dead time of the SPD based on the Si-APD.
The number of the trapped carriers that cause the afterpulsing can be decreased by reducing the avalanche gain of the APD, because the low avalanche gain doesn’t produce a large current passing through the APD. However, it is difficult to discriminate a small avalanche signal from a charge pulse through APDs. Here, the charge pulse (10~20 mV at 50Ω termination) is a transferred signal of the rectangular bias pulse for the gating operation, which is due to the capacitor like response of the APDs. The discrimination level can be reduced to less than 10 mV by the discharge measurement  or use of the charge-pulse cancellation circuits [5, 6]. However, it is difficult to reduce the discrimination level to less than 1mV using these methods. Recently we have proposed and demonstrated a sine wave gating scheme which allows the discrimination of an extremely small avalanche signal without being disturbed by the charge pulses. The preliminary results indicated the possibility of a high speed gating operation . In this paper, we report on an 800MHz operation of the SPDs using the sine wave gating.
2. InGaAs/InP avalanche photodiode operated with a sine wave gating
The diagram of our single-photon detection scheme is depicted in Fig. 1(a). In order to supply an AC voltage (gate signal) conjugated with a DC reverse bias voltage VDC to the InGaAs/InP-APD, we used the gated passive quenching circuit (GPQC) . The gate signal is a sine wave with a frequency of ωg. In the case that an avalanche is not triggered in the APD, only the gate signal is outputted from the GPQC. Here, the amplitude of the sine wave signal is strongly attenuated by the APD’s resistance which is much lager than the resistance RO. The transferred gate signal disturbs the discrimination of the avalanche signal. However, the signal can be easily removed by the band elimination filter (BEF) with the center frequency of ωg. On the other hand, an electric energy of the avalanche signal is distributed to many frequency components, because the gating operation makes the waveform of the avalanche signal impulse-like. Therefore almost all the energy of the avalanche signal can be passed through the BEF. As a result, the AC noise-free discrimination of avalanche signals can be realized even if the APD is operated with the gated mode. Here, when a BEF with an extremely narrow band width is employed and the impedance is perfectly matched, the efficiency that the electric circuit transfers the energy of the avalanche signals is close to unity and the maximum signal-to-noise ratio can be obtained. Figure 1(b) shows the measured spectra of the GPQC outputs without the BEF when the 250MHz sine wave signal with the peak-to-peak amplitude of 8Vp-p was applied for the gate. Here, although the large amplitude of the gate voltage is not required for getting the avalanche signals, it is beneficial to reduce the voltage across the APD during the gate-off time. The lower voltage shortens the lifetime of the trapped carriers . The APD was operated at room temperature and the breakdown voltage VB was 52.1V. First, the VDC supplied to the APD was set to 50V and the excess voltage VE above the VB was 1.9V. Since few avalanche events occurred, we could measure only the transfer response of the GPQC against the gate signal input. The energy of the transferred signal concentrated at the frequency components of ωg, 2ωg, and 3ωg. The higher order harmonics (2ωg and 3ωg) were generated by a nonlinear diode response of the APD. On the other hand, when VDC was set to 52.3V (VE=4.2V), the GPQC outputted not only the transferred gate signal but also additional signals consisting of many frequency components. The additional signals are due to the avalanche events caused by thermal carriers. The BEFs have the power rejection ratio of 66dB (bandwidth~100MHz) and 38dB (bandwidth~150MHz) at the center frequency of 250MHz (ωg) and 500MHz (2ωg), respectively. Figure 1(c) shows the measured avalanche signal that was amplified by 20dB after passing through the BEFs. The peak amplitude of the avalanche signal was approximately 1 mV without the amplification. Then the noise level was ~0.1mV and the discrimination level was set to 0.5mV.
3. Experimental setup
The experimental setup is shown in Fig. 2. InGaAs/InP-APD (EPITAXX EPM239BA) was electrically cooled at -35 degrees Celsius. The gate signal (2Vp-p) was generated by the signal generator (SG). The 10% of the signal was used to trigger the laser pulse. The 90% of the signal was amplified to 12Vp-p (10Vp-p at 800MHz), and was used for gating. The avalanche signals passing through the BEF were amplified by 20dB and subsequently led to the discriminator. The setup depicted in Fig. 2(a) was used for the measurement of the quantum efficiency. The trigger rate for the laser was set to wg/64 (ωg/128 at 800MHz) by the prescaler, because the maximum repetition rate of the laser was 10MHz. The DFB-laser emitted the 50-ps optical pulses at 1550-nm. They were attenuated to the single-photon level (0.1-photon/pulse) by the optical attenuator (AT) and led to the APD through the single-mode fiber (SMF). On the other hand, using the setup depicted in Fig. 2(b), the autocorrelation of the dark counts was measured to evaluate the afterpulsing probability . The measurement was carried out using the multi channel scaler (MCS). The digital signals were divided into two signals that were used for the start and the stop signals of the MCS, respectively. If the MCS is triggered by the start signal (an initial dark count), it records the arrival time of the following stop signals with the time resolution of 2.5 ns (bin). Here, the measurement duration was 163.8 μs in which there exists 216 bins. Since the MCS starts recording 50 ns after it is triggered by the start signal, the stop signals were delayed using the additional 10 m long cable. The measured histogram of time at which the dark counts and the afterpulses occurred can be expressed as the autocorrelation function,
where k labels the bin, p(0|k) is the probability seeing an related afterpulse at the k-th bin (after 2.5×k ns) subsequent to the initial dark count, and n̄ is the mean number of the counts per bin. The overall afterpulsing probability Pa is given by
The initial stop signal indicates an avalanche event caused by thermal carriers and is recorded at k=0 bin. Then the MCS stops recording the following stop signal for 50 ns, namely, the signal-pair resolution of the MCS is 50 ns. Therefore, we can not experimentally measure the afterpulsing probability up to 20 bins (50 ns). Hence, we calculated the afterpulsing probability using the single-exponential-decay function α+βe -γk (α, β, and γ were arbitrary coefficients). Note that the measured dark count probability is very small (10-5), the probability that more than two events occure in the 50 ns can be negligible.
We investigated the quantum efficiency, the dark count probability, and the afterpulsing probability of the APD changing the excess bias voltage VE and the frequency wg (100~800MHz) of the sine-wave gate signal. Here, the VE was adjusted by changing the VDC. The quantum efficiencies at different frequencies (ωg) were shown as a function of the VE in Fig. 3. As the ωg becomes higher, the gate-on time Tg (see the illustration incorporated with Fig. 3) becomes shorter when we keep the VE constant. If the Tg is too short, the detectable avalanche signals cannot be obtained because of the APD’s optical response time of a few hundreds picoseconds . However, increasing the VE makes the optical response time shorter and the Tg wider. As a result, the high quantum efficiency was obtained despite the high ωg. The short Tg is valuable for reducing the number of avalanche carriers passing through the APD, because the avalanche is quickly quenched. However, when the ωg is too high, the afterpulsing probability increases, because a large numbers of gates are applied during the lifetime of the trapped carriers. This is the reason why there exists a ωg that minimizes the afterpulsing probability. Figure 4(a) shows the relation between the quantum efficiency and the afterpulsing probability. When the ωg was 500MHz at which the minimum noise equivalent power was obtaind, the afterpulsing probability was only 6.5% with the quantum efficiency of 13% and the dark count probability per gate of 1.0×10-5 (see Fig. 4(b)). Moreover, when the ωg was 800MHz and the temperature of the APD was set to -30 degrees Celsius, the afterpulsing probability was only 6.0% with the quantum efficiency of 8.5% and the dark count probability per gate of 9.2×10-6.
5. Advantage of the sine wave gating
The afterpulsing probability strongly depends on the number of trapped carriers. In order to reduce the afterpulsing probability, the number of multiplied carriers passing through the APD must be reduced by shortening the gate duration (quenching time) and reducing the excess voltage. However, in the case of the square pulse gating, the transferred gate signal (charge pulse) disturbs discrimination of the small avalanche signal. Although the charge pulse can be
reduced by the special cancellation circuit, the residual noise level is not so low because of the difficulty of balancing the signals. The sine wave gating scheme can discriminate the small avalanche signal because the transferred gate signal is removed by the BEF. In our experiments, the excess voltage was set to lower than that of the conventional square pulse gating when the gate frequency is less than 250 MHz. On the other hand, when the gate frequency is higher than 500 MHz, the avalanche cannot be triggered with the low excess voltage because the gate duration is close to the optical response time of the APD. Taking into account the frequency and amplitude of the sine wave gate and the excess voltage, the gate duration is shorter than 500 ps. In such a case, the excess voltage must be set to higher value to obtaine the small avalanche signal that contains ~105 electrons. then the discrimination level is 2~3 orders of magnitude lower than that of the conventional square pulse gating scheme, which results in the strong suppression of the afterpulsing.
Table 1 shows the comparison between our results and the other reported ones for high speed single-photon detection. Our results outperform the other groups’ ones remarkably, although the afterpulsing probability is somewhat high. In the case of 500MHz operation, the photon-counting rate will be 6.5×106 when the average incident photon number is 0.1 per pulse. On the other hand, in the case of 800MHz operation, the photon-counting rate will be 6.8×106. The difference of the count rates is not so high. However, the afterpulsing probability and the dark count probability per gate at 800MHz are smaller than those at 500MHz.
we have investigated single-photon detection at 1550-nm using the InGaAs/InP-APD operated with the sine wave gating. The gating scheme enables us to discriminate the avalanche signal which is one or two orders of magnitude smaller than that obtained in other schemes. As a result, the repetition frequency of the gate can be reached 800MHz with a low afterpulsing probability. By applying our high-speed SPD to a QKD system, the bit rate of the QKD would be dramatically increased.
This research was partially supported by the Ministry of Education, Science, Sports and Culture, Grant-in-Aid for Scientific Research on Priority Areas, 17068001.
References and links
1. D. Stucki, N. Gisin, O. Guinnared, G. Ribordy, and H. Zbinden, “Quantum key distribution over 67 km with a plug & play system,” New J. Phys. 4, 41, (2002) [CrossRef]
2. G. Ribordy, N. Gisin, O. Guinnard, D. Stucki, M. Wegmuller, and H. Zbinden, “Photon counting at telecom wavelengths with commercial In-GaAs/InP avalanche photodiodes: Current performance,” J. Mod. Opt. 51, 1381–1398, (2004)
3. N. Namekata, Y. Makino, and S. Inoue, “Single-photon detector for long-distance fiber-optic quantum key distribution,” Opt. Lett. 27, 954–956, (2002) [CrossRef]
4. A. Yoshizawa, R. Kaji, and H. Tsuchida, “Gated-mode single-photon detection at 1550 nm by discharge pulse counting,” Appl. Phys. Lett. 84, 3606–3608 (2004) [CrossRef]
5. D. S. Bethune, W. P. Risk, and G. W. Pabst, “A high-performance integrated single-photon detector for telecom wavelengths,” J. Mod. Opt. 51, 1359–1368, (2004)
6. A. Tomita and K. Nakamura, “Balanced, gated-mode photon detector for quantum-bit discrimination at 1550 nm,” Opt. Lett. 27, 1827–1829 (2002) [CrossRef]
7. C. Langrock, E. Diamanti, R. V. Roussev, Y. Yamamoto, and M. M. Fejer, “Highly efficient single-photon detection at communication wavelengths by use of upconversion in reverse-proton-exchanged periodically poled LiNbO3 waveguides,” Opt. Lett. 30, 1725–1727, (2005) [CrossRef] [PubMed]
8. S. Sasamori, N. Namekata, and S. Inoue, IEICE trans.Japanese ed., to be published (2006)
10. A. Yoshizawa, R. kaji, and H. Tsuchida, “10.5 km fiber-optic quantum key distribution at 1550 nm with a key rate of 45 kHz,” Jpn. J. Appl. Phys. bf43, 735–737, (2004) [CrossRef]
11. P L. Voss, K G. Köprülü, S. Choi, S. Dugan, and P. Kumar, “14-MHz rate photon counting with room temperature InGaAs/InP avalanche photodiodes,” J. Mod Opt. 51, 1369–1379 (2004)