Superconducting Nanowire Single-Photon Detector With ... - IEEE Xplore

Superconducting Nanowire Single-Photon Detector With a System Detection Efficiency Over 80% at 940-nm Wavelength Volume 8, Number 2, April 2016 W. J. Zhang H. Li L. X. You J. Huang Y. H. He L. Zhang X. Y. Liu S. J. Chen Z. Wang X. M. Xie

DOI: 10.1109/JPHOT.2016.2542838 1943-0655 Ó 2016 IEEE

IEEE Photonics Journal

SNSPD With SDE Over 80% at 940-nm Wavelength

Superconducting Nanowire Single-Photon Detector With a System Detection Efficiency Over 80% at 940-nm Wavelength W. J. Zhang, H. Li, L. X. You, J. Huang, Y. H. He, L. Zhang, X. Y. Liu, S. J. Chen, Z. Wang, and X. M. Xie State Key Laboratory of Functional Materials for Informatics, Shanghai Institute of Microsystem and Information Technology (SIMIT), Chinese Academy of Sciences (CAS), Beijing 100190, China DOI: 10.1109/JPHOT.2016.2542838 1943-0655 Ó 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

Manuscript received February 3, 2016; revised March 7, 2016; accepted March 10, 2016. Date of publication March 16, 2016; date of current version April 4, 2016. This work was supported by the National Natural Science Foundation of China under Grant 61401441 and Grant 61401443 and in part by the Strategic Priority Research Program (B) of the CAS under Grant XDB04010200 and Grant XDB04020100. Corresponding author: L. X. You (e-mail: [email protected]).

Abstract: Implementations of quantum information require single-photon detectors (SPDs) with high detection efficiency (DE) at a wavelength of 940 nm, which is a challenge for the available semiconducting SPDs. Superconducting nanowire SPDs (SNSPDs) are capable of detecting visible and near-infrared single photons with high DE. However, these detection capabilities place stringent design requirements on the cavity and nanowire geometry structures. We design, fabricate, and measure SNSPDs with high DE optimized for the 940-nm wavelength. The NbN SNSPDs were fabricated on 1-D photonic crystals for high optical absorptance. By tuning the filling factor of the nanowire through numerical simulations and experiments, we were able to obtain an SNSPD (7 nm thick, 125 nm width, and 0.57 filling factor, as well as active area of 18 18 m) with a saturated system DE of 83.6 3.7%, at a dark count rate of 10 Hz, and a low polarization dependence of 1.17 0.02. To our best knowledge, this is the highest value reported for NbN SNSPDs at 940-nm wavelength. The availability of an SNSPD with high system DE at 940 nm may have a profound impact in the field of photonic quantum technologies, such as multiphoton entanglement. Index Terms: Niobium nitride (NbN), superconducting nanowire single-photon detectors (SNSPDs), system detection efficiency, quantum information.

1. Introduction The single-photon detector (SPD) plays a key role in many scientific fields related to photon detection, such as quantum information (QI) science [1], [2]. Performance parameters such as the detection efficiency (DE) of the SPD are crucial to several advanced scientific processes, such as the 200 km demonstration of measurement-device-independent quantum key distribution [3], photon entanglement [4], and linear optical computation [5]. Implementations of QI such as multiphoton entanglement often require high efficiency SPDs operating at a wavelength of 940 nm, since the best available quantum-dot single-photon sources operate at this wavelength [6], [7]. However, the best available Si SPD typically has a DE of approximately 20%, which limits the measurement of the number of entangled photons. The superconducting nanowire single-photon detector (SNSPD) [8] is a promising candidate for single-photon

Vol. 8, No. 2, April 2016

4500908



detection, owing to its high system detection efficiency (SDE), high speed, low dark count rate (DCR), and small timing jitter. Most previous of SNSPDs have focused on a 1550 nm wavelength [9], [10]. Recently, research has suggested that SNSPDs may also have excellent performance at shorter wavelengths [11]–[13]. Since the SDE of SNSPD can be expressed as SDE ¼ AE OCE IDE [9], where AE is the absorption efficiency in nanowires, OCE is the optical coupling efficiency between the incident photons and the active area, and IDE is the intrinsic detection efficiency or pulse generation probability of the nanowire. Therefore, it is complicated to achieve high SDE, which requires all the factors improved. In previous work, we showed a non-saturated SDE of 78% at 850-nm wavelength for the large active area SNSPDs coupling with a multimode fiber [12]. We also demonstrated low-filling-factor SNSPDs with an SDE of 60% and a recovery time of 12 ns at a wavelength of 940 nm [13]. However, further improve the SDE of the NbN-type SNSPD is necessary. It is desirable to find out the limitations that prevent these detectors to achieve near-unity SDE. In this research, we reported SNSPDs with a high SDE of over 80% by optimizing optimize AE, OCE, and IDE systematically. The SNSPDs were made of 7-nm-thick NbN film grown on an Si substrate using 13 periodic Ta2 O5 =SiO2 bilayers as 1-D photonic crystals (1-D PC) to achieve high optical absorptance. The nanowire had a nominal width of 125 nm, and the filling factor was tuned, allowing the dependence of the AE on the filling factor to be systematically examined through numerical simulations and experiments. In order to enable unity OCE, the active area of SNSPDs was designed to 18 18 m2 . Uniform nanowires were fabricated to ensure high IDE. The best detector showed a saturated SDE of 83.6 3.7% for the wavelength of 940 nm, with a DCR of 10 Hz, and a low polarization dependence of 1.17 0.02. This is, to the best of our knowledge, the highest SDE reported for the NbN-type SNSPDs at 940 nm wavelength.

2. Device Design and Fabrication The design of SNSPDs for 940 nm follows the principle of integrating 1-D PC structure to enhance the absorption of the nanowires [11]–[14], as shown in Fig. 1(a). The 1-D PC comprised periodic SiO2 and Ta2 O5 bilayers and worked as a quarter wavelength resonator, resulting in the enhancement of optical absorption of the nanowire for the target wavelength. The thicknesses of the bilayers were designed to be a quarter of the 940 nm wavelength, i.e., 104 nm for the Ta2 O5 layer and 160 nm for the SiO2 layer. Thirteen periodic SiO2 =Ta2 O5 bilayers were alternately deposited on the Si substrate using ion beam sputtering, and the film thickness was monitored. The measured reflectivity of the PC substrate was greater than 99% at 940 nm, and the root-mean-square roughness of the NbN thin film was less than 0.2 nm. A 7-nm-thick NbN film was then deposited onto the substrate at room temperature using reactive direct-current magnetron sputtering in an Ar-N2 gas mixture with partial pressures of 79% and 21%, respectively. The thickness of the film was controlled by the sputtering time. After that, the NbN film was patterned into a meandered nanowire structure using electron beam lithography (EBL) with a positive-tone polymethyl methacrylate electron-beam resist and was reactively etched in CF4 plasma. Finally, a 50- -matched coplanar waveguide was formed using ultraviolet lithography and reactive ion etching. In our previous work, we demonstrated that good optical coupling was obtained in an active area with a diameter 15 m, and strong saturation dependence of the SDE on the bias current was observed at a wire width of 110 nm [3]. Thus, to optimize the SDE, the new SNSPDs were designed to have a sensitive area with a lateral width of 18 m, which is much larger than the 5.6 m mode-field diameter of a standard single-mode fiber for 940 nm (SM-800). The nanowire linewidth was set at 125 nm to reduce the kinetic inductance (KI) and polarization dependence. To optimize the optical absorptance of the nanowires, we prepared devices with spacings of 75, 95, 115, 135, 155, and 265 nm (filling factors of 0.32–0.63). A scanning electron microscopy (SEM) image of the nanowire with 125 nm width and 95 nm spacing nanowire is shown in Fig. 1(b), demonstrating the precision of the fabrication process.


4500908



Fig. 1. (a) Schematic of the 7-nm-thick NbN SNSPD fabricated on a PC substrate, coupled with an anti-reflection-coating single-mode fiber for 940 nm. The PC substrate comprised 13 periodic Ta2 O5 =SiO2 bilayers deposited on a silicon wafer with 99% reflectance measured at a wavelength of 940 nm. (b) SEM image of nanowires. The wires are 125 nm wide with a 95 nm spacing. (c) Calculated parallel polarized absorptance (A== , left axis) and PER(A== =A? , right axis) as functions of spacing for nanowires with fixed width of 125 nm and thicknesses of 7, 6, and 5 nm, using the rigorous coupled-wave analysis. The red (blue) arrow shows that the peak of A== (lower PERcal ) moved to a smaller spacing as the thickness decreased. (d) Wavelength dependence of A== and A? for the nanowire with 7 nm thickness, 125 nm width, and 95 nm spacing. (e) Schematic of experimental setup for SNSPDs measurement. The optical and electrical paths are indicated with red and blue lines, respectively.

Fig. 1(c) shows the parallel polarized absorptance ðA== Þ and polarization extinction ratio (PER) dependence on the spacings for 940-nm-wavelength incident light. We calculated the absorptance for the nanowire with a fixed nanowire linewidth of 125 nm, and varied thickness from 5 to 7 nm using rigorous coupled-wave analysis [12]. For the 7-nm-thick nanowires (blue solid-line), A== showed a broad peak with values of 99–100% at a spacing of 100–170 nm, corresponding to a filling factor of 0.56–0.42. When the spacing between the nanowires was greater than 135 nm, the optical absorptance monotonically decreased, reaching 92% at a spacing of 300 nm. As the


4500908



Fig. 2. (a) Parallel-polarized SDE ðSDE== Þ for 940 nm vs. the normalized bias current Ib =Isw for SNSPDs with various filling factors (0.32–0.64) at 2.3 K. The solid lines were empirically fitted with sigmoid functions. The experimental data were named as width-spacing-filling factor. (b) DCRs vs. I b =I sw corresponding to the devices shown in Fig. 2(a), with the same symbols. The DCRs were measured with unshielded bare fiber attached at 300 K. (c) DCRs vs. Ib =Isw for a device with a filling factor of 0.57, measured under three different conditions. 1) SNSPD was well shielded at 2.3 K with no attached fiber (DCRi shown as black squares). 2) The input bare fiber at 300 K was shielded in an aluminum box to remove the influence of the stray light (shown as violet triangles). 3) The input bare fiber at 300 K was unshielded and exposed to the room temperature blackbody radiation. The results shown in Fig. 2(b) were also measured at this condition.

thickness decreased, the peak values of A== shifted to the small spacing (high filling factor) structure. The calculated PER (PERcal , defined as A== =A? ) gradually decreased as the spacing and thickness decreased, with a range from 1.83 to 1.26. Fig. 1(d) shows the calculated wavelength dependence of A== and A? for a 7-nm-thick, 125-nm-wide and 95-nm-spacing nanowire. The resonated bandwidth ranged from 820 to 1090 nm and was defined by a 3 dB cutoff.

3. Device Performance Fig. 1(e) demonstrates our experimental setup for SNSPDs measurement. First, the devices were mounted in a fiber-coupled package optimized for front-side illumination with an antireflection coated single-mode fiber (Thorlabs, SM-800). Then, the packaged devices were loaded in a cryostat and cooled to 2.3 K in a 6-channel closed-cycle Gifford-McMahon (GM) cryocooler. Each package was connected to a semi-rigid coaxial cable and coupled with a single-mode optical fiber. Outside the cryocooler, the input and output signals were transmitted through a bias tee, with the semi-rigid coaxial cables connecting. We counted electrical pulses from the detectors using room a temperature low-noise amplifiers (RF Bay Inc., LNA650, 50 dB gain) and a pulse counter (SRS Inc., SR400). To optically probe the devices, we used a 940 nm continuous-wave fiber laser that was attenuated and sent into the cryostat via a SM-800 fiber. A fiber polarization controller was connected after the attenuators to maximize detector counts. Before measurements were taken, we carefully calibrated the attenuators for the 940 nm wavelength using a power meter (Agilent 81624B). The light was then attenuated to a photon flux of 0.1 M photons/s. The fiber was cut and spliced with the one connected to the detector in the cryostat. The light power was calibrated using an optical fiber without anti-reflection coating, which would result in an overestimation of the SDE by approximately 4% because of Fresnel reflection. Thus, the measured SDEs were corrected for this effect by a dividing factor of 1.04. Fig. 2 shows the SDE== and DCR as functions of the normalized bias current for SNSPDs with various nanowire spacings. Solid lines show the empirical fitting results using sigmoid functions for the SDE== data. The experimental data were named as width-spacing-filling factor. The measured linewidths had an of approximately ±4 nm in the designed value, as a result of the


4500908



limited accuracy of SEM measurements and the fabrication consistency [see Fig. 1(b)]. The switching currents ðIsw Þ of these meanders ranged from 17 to 23 A, and the transition temperature ðTc Þ were approximately 8.2 K. As the filling factor increased from 0.32 to 0.64, the SDE== s first increased and then fell, consistent with the trend of calculation. Thus, the optimization of the filling ratio is necessary. The highest SDE, 83.6%, was obtained in the structure denoted “125-95-0.57,” which had a filling factor of approximately 0.57. Since the SDE included all the losses in the system, the uncertainty associated with the SDE values was estimated to be less than 3.7% [13], which arose mainly from the uncertainty of the optical power calibration. An upturn in the SDE curve was observed at bias current close to Isw ðIb =Isw > 0:96Þ, caused by the abnormal increase of dark counts under illumination [15]. To avoid misunderstanding, we excluded the data for the bias currents close to Isw . Fig. 2(b) shows the DCRs as functions of the normalized bias current ðIb =Isw Þ. These DCRs were measured with unshielded bare fiber attached at 300 K. The DCR at Ib =Isw G 0:88 was mainly contributed by the background DCR ðDCRBG Þ, while that at Ib =Isw > 0:88 was mainly a result of the intrinsic DCR ðDCRi Þ, originating from the vortex-antivortex motion [16]. In our experiment, we measured the DCR by controlling the shutter of the attenuator. There were two sources of DCRBG : blackbody irradiation of the fiber attached to SNSPDs [17] and the stray light entering the bare fiber that was used in the measurements. To closely investigate the system DCR caused by the fiber blackbody radiation, we carefully examined the DCRs of the device named “125-95-0.57,” under three different condition. These are shown as the three curves in Fig. 2(c). In condition 1, the SNSPD was well shielded at 2.3 K without an attached fiber. In condition 2, the input bare fiber at 300 K was shielded by an aluminum box to remove the influence of the stray light. In condition 3, the measurements were taken under the same conditions as those for obtaining the results shown in Fig. 2(b). The results suggested that DCRi can be suppressed to less than 0.01 Hz when the bias current is lower than 0:85Isw . Thus, the stray light effectively contributed to the background DCR in Fig. 2(b). SNSPDs can therefore be operated with a DCRBG at 1 Hz or 0.1 Hz, provided a customized armored SM-800 fiber is used. In particular, at approximately 1 Hz, the DCRBG of SNSPDs for 940 nm was nearly two orders of magnitude smaller than the SNSPDs for 1550 nm, which are generally of the order of 100 Hz [10], [15]. There are three reasons for this phenomenon. First, the fiber blackbody radiation at ≤ 300 K is less at 940 nm wavelength than at 1550 nm. Second, a larger nanowire linewidth corresponds to lower sensitivity to the photons with longer wavelength. Third, the PC was optimized at 940 nm with a bandwidth of approximately 270 nm, which suppressed the absorption of the blackbody radiation as well as the stray light at other wavelengths. We defined the SDE==max and SDE?max as the SDE for the parallel and perpendicular polarization at a DCR of 100 Hz, respectively. The SDE==max were nearly the same as the asymptotic SDE ðSDEasy Þ evaluated from fitting by a sigmoid function in Fig. 2(a). However, because of non-saturation in the SDE, an exception appeared for the SNSPD named, “125-75-0.63,” which had an SDE==max of 76.7% at the DCR of 100 Hz, and an SDEasy of 80.3%. This SDEasy was plotted with a blue open star in Fig. 3(a) and it is more appropriate to estimate the absorption for the non-saturated device [10]. As shown by the above calculations, A== as a function of spacing reached maxima at a certain range of spacing. In order to compare the experimental data with the simulation results, an optimized fitting curve (black solid line) was plotted by multiplying a factor of 0.82 to A== as shown in Fig. 3(a). The peak values of measured SDE appear at spacing of approximately 85–115 nm, where seems a little smaller than the simulated results show. A considerable discrepancy between the simulations and measurements appeared at the spacing of 265 nm, which was possibly caused by the reduction of absorption of nanowire formed in fabrication process. Although the observed nanowire linewidth is consistent with the design, the nanowire cross-section showed a trapezoidal shape revealed by TEM [18]. Since the nanowire cross-section is mainly determined by the cross-section of photoresist and etching process. The trapezoidal cross-section for the positive-tone resist PMMA would be more significant, when a larger exposed area (i.e. spacing) between the nanowires exists. Thus, the over-etching on


4500908



Fig. 3. (a) Measured SDE==max (red dots), simulated A== (red dashed-line), and fitted absorption (A== 0:82, black solid line) as functions of spacing. The blue open star indicated the asymptotic SDE for device of 125-75-0.63. (b) Measured (red dots), calculated (blue dashed-line), and fitted (black solid-line) PERs as functions of spacing. The experimental spacing ranged from 75 to 265 nm, corresponding to filling factors from 0.32 to 0.63. The fitted data was obtained from the calculated data for a 7-nm-thick nanowire by adopting a reduced factor of 0.80.

sidewall could cause the reduction of absorption and the decrease of SDE at large spacing. Optimizing the fabrication process would reduce this effect and improve the SDE close to the simulation value. Fig. 3(b) shows the measured (red dots) and calculated (blue dashed-line) PERs ðSDE==max = SDE?max Þ as functions of spacing. The relative error for the measured PER was estimated to be less than 3%. All the measured values were lower than the calculated ones. In order to fit the experimental data, a reduction factor of 0.80 was applied and shown as black-solid line in Fig. 3. The difference between experimental and calculated values may be due to imperfect nanowire cross-sections mentioned above or variation in the reflective indices of materials at low temperatures. Another possible explanation is the unknown state of polarization of the fiber at low temperature. More work is necessary to understand these results. Fig. 4 shows that the SNSPD with the best SDE named as “125-95-0.57.” SDE== and SDE? as functions of the normalized bias current are shown in Fig. 4(a), giving a low PER of 1.17 0.02. This PER is comparable with the one reported for WSi SNSPDs [9]. Fig. 4(b) shows the SDE== as a function of DCR with a shielded fiber attached. Clearly, the SDE was over 70% (80%) at a DCR of 0.1 Hz (1 Hz), and saturated at a DCR of 10 Hz, with a value of 83.6%. Because of the relatively large active area of these devices, the recovery time was limited by the large KI of the nanowire [19]. As shown in Fig. 4(c), a recovery time of approximately 27.8 ns was deduced from the decay time of the respond pulse, which was defined as 1/e (36.8%) of the maximum pulse amplitude. Fig. 4(d) demonstrated the timing jitter of the SNSPD measured at a bias current of 19.1 A (0.91 Isw ) using a time-correlated single-photon counting (TCSPC) module (SPC-150, Becker & Hickl GmbH). A 1550-nm-wavelength pulsed laser with a 100 fs pulse width was used, since no fs-pulsed laser at 940 nm was available. The input photon flux was also maintained in the single-photon level. The timing jitter defined by the full-width at halfmaximum of the histogram was 62.7 ps.

4. Discussion The maximal SDEs of over 70% (80%) at a DCR of 0.1 Hz (1 Hz) for 940 nm wavelength are sufficiently high to support many applications. This may have a profound impact in the field of photonic quantum technologies. So far, the maximum number of successfully entangled photons remains at eight [10], however, ten-photon entanglement still remains a challenge. The most promising approach to tackle this problem is the use of semiconductor quantum dots that emit single photons at the wavelengths of approximately 940 nm. Given ten such single-photon sources with a repetition rate of 10 MHz, we can estimate a ten-photon coincidence count rate


4500908



Fig. 4. (a) SDE== and SDE? vs. Ib =Isw for the device with 125 nm width and 95 nm spacing, giving a low PER of 1.17. Solid lines were empirically fitted with sigmoid functions. The switching current of the detector was 20.8 A. (b) SDE== and SDE? vs. DCR. (c) Real time and averaged (red curve) oscilloscope traces of the response pulse. The vertical arrow indicates the recovery time (27.8 ns) defined as 1/e (36.8%) of the maximum pulse amplitude. (d) Timing jitter measured at 1550 nm via a fs-pulsed laser. The red line is the Gaussian function fitting of the measured histograms of the time-correlated counts. The timing jitter was 62.7 ps, measured at a bias current of 19.1 A (0.91 Isw ).

of 1 Hz using a conventional Si SPD with a DE of 20%. Using the newly developed 940 nm SNSPD, the ten-fold count rate will increase to 1 MHz—a marked six orders of magnitude improvement. However, 16% photons were missing in the detection, including a 5–6% reflection on the surface of nanowires, and an imperfect AE provide by the non-ideal PC structure. Further improvement of the SDE will require the optimization of the optical stack design. The use of antireflective coating on the surface of NbN nanowires may further reduce reflection and the PER, and possibly improve the measured SDE. Another reason could be the limited IDE of NbN material, although we assumed the saturated SDE indicated a 100% IDE. Further works are needed to realize an SDE close to unity. On the other hand, reduction of the recovery time is potentially interesting in QI applications. A direct approach would be to reduce the active area whilst maintaining an appropriate filling factor; slightly increasing of the nanowire linewidth may also be helpful. Another approach would be to adopt the array of SNSPDs [20]. Overall, we believe that the performance of the SNSPD can be further improved in the near future.

5. Conclusion We designed, fabricated and measured NbN SNSPDs with high DE optimized for the 940 nm wavelength. Using 1-D PC and tuning the filling ratio of the nanowire through simulations and experiments, we obtained an SNSPD (7 nm thick, 125 nm width, and 0.57 filling factor, as well as an active area of 18 18 m) with a saturated system DE of 83.6 3.7% at a DCR of 10 Hz


4500908



and a low polarization dependence of 1.17 0.02. To our best knowledge, this is the highest value reported for NbN SNSPDs at 940 nm wavelength. The availability of an SNSPD with high system DE at 940 nm may have a profound impact in the field of photonic quantum technologies, such as multiphoton entanglement.

Acknowledgment The authors would like to thank Prof. C. Y. Lu from the University of Science and Technology of China for valuable discussions.

References [1] C. Y. Lu and J. W. Pan, “Quantum optics: Push-button photon entanglement,” Nature Photon., vol. 8, pp. 174–176, Feb. 2014. [2] R. H. Hadfield, “Single-photon detectors for optical quantum information applications,” Nature Photon., vol. 3, pp. 696–705, Nov. 2009. [3] Y. L. Tang et al., “Measurement-device-independent quantum key distribution over 200 km,” Phys. Rev. Lett., vol. 113, no. 19, Nov. 2014, Art. no. 190501. [4] X. C. Yao et al., “Observation of eight-photon entanglement,” Nature Photon., vol. 6, pp. 225–228, Feb. 2012. [5] E. Knill, R. Laflamme, and G. J. Milburn, “A scheme for efficient quantum computation with linear optics,” Nature, vol. 409, pp. 46–52, Jan. 2001. [6] Y. J. Wei et al., “Deterministic and robust generation of single photons from a single quantum dot with 99.5% indistinguishability using adiabatic rapid passage,” Nano Lett., vol. 14, no. 11, pp. 6515–6519, Nov. 2014. [7] W. B. Gao, P. Fallahi, E. Togan, J. Miguel-Sanchez, and A. Imamoglu, “Observation of entanglement between a quantum dot spin and a single photon,” Nature, vol. 491, pp. 426–430, Nov. 2012. [8] C. M. Natarajan, M. G. Tanner, and R. H. Hadfield, “Superconducting nanowire single-photon detectors: Physics and applications,” Supercond. Sci. Tech., vol. 25, no. 6, Apr. 2012, Art. no. 063001. [9] F. Marsili et al., “Detecting single infrared photons with 93% system efficiency,” Nature Photon., vol. 7, pp. 210–214, Feb. 2013. [10] T. Yamashita, S. Miki, H. Terai, and Z. Wang, “Low-filling-factor superconducting single photon detector with high system detection efficiency,” Opt. Exp., vol. 21, no. 22, pp. 27177–27184, Nov. 2013. [11] D. K. Liu et al., “Multimode fiber-coupled superconducting nanowire single-photon detector with 70% system efficiency at visible wavelength,” Opt. Exp., vol. 22, no. 18, pp. 21167–21174, Sep. 2014. [12] H. Li et al., “Large-sensitive-area superconducting nanowire single-photon detector at 850 nm with high detection efficiency,” Opt. Exp., vol. 23, no. 13, pp. 17301–17308, Jun. 2015. [13] W. J. Zhang et al., “Superconducting nanowire single-photon detectors at a wavelength of 940 nm,” AIP Adv., vol. 5, no. 6, Jun. 2015, Art. no. 067129. [14] B. Baek, J. A. Stern, and S. W. Nam, “Superconducting nanowire single-photon detector in an optical cavity for front-side illumination,” Appl. Phys. Lett., vol. 95, no. 19, Nov. 2009, Art. no. 191110. [15] S. J. Chen et al., “Dark counts of superconducting nanowire single-photon detector under illumination,” Opt. Exp., vol. 23, no. 8, pp. 10786–10793, Apr. 2015. [16] H. Bartolf et al., “Current-assisted thermally activated flux liberation in ultrathin nanopatterned NbN superconducting meander structures,” Phys. Rev. B, vol. 81, Jun. 2010, Art. no. 024502. [17] X. Y. Yang et al., “Superconducting nanowire single photon detector with on-chip bandpass filter,” Opt. Exp., vol. 22, no. 13, pp. 16267–16272, Jun. 2014. [18] H. Li et al., “Nonideal optical cavity structure of superconducting nanowire single photon detector,” IEEE J. Sel. Topics Quantum Electron., vol. 20, no. 6, Nov./Dec. 2014, Art. no. 3803705. [19] A. J. Kerman et al., “Kinetic-inductance-limited reset time of superconducting nanowire photon counters,” Appl. Phys. Lett., vol. 88, no. 11, Mar. 2006, Art. no. 111116. [20] D. Rosenberg, A. J. Kerman, R. J. Molnar, and E. A. Dauler, “High-speed and high-efficiency superconducting nanowire single photon detector array,” Opt. Exp., vol. 21, no. 2, pp. 1440–1447, Jan. 2013.


4500908