NbN nanowire superconducting single-photon detector for mid-infrared

Transcription

1 Available online at Physics Procedia 36 (2012 ) Superconductivity Centennial Conference NbN nanowire superconducting single-photon detector for mid-infrared A. Korneev, Yu. Korneeva, I. Florya, B. Voronov, G. Goltsman Moscow State Pedagogical University, 1 Malaya Pirogovskaya, Moscow , Russia Abstract Superconducting single-photon detectors (SSPD) is typically 100 nm-wide supercondiucting strip in a shape of meander made of 4-nm-thick film. To reduce response time and increase voltage response a parallel connection of the strips was proposed. Recently we demonstrated that reduction of the strip width improves the quantum efficiency of such a detector at wavelengths longer than 1.5 μm. Being encourage by this progress in quantum efficiency we improved the fabrication process and made parallel-wire SSPD with 40-nm-wide strips covering total area of 10 μm x 10 μm. In this paper we present the results of the characterization of such a parallel-wire SSPD at 10.6 μm wavelength and demonstrate linear dependence of the count rate on the light power as it should be in case of single-photon response. c Published by by Elsevier Ltd. B.V. Selection Selection and/or and/or peer-review peer-review under responsibility under responsibility of Horst Rogalla of the and Guest Peter Editors. Kes. Open access under CC BY-NC-ND license. Keywords: Infrared single-photon detectors, superconducting device fabrication, superconducting NbN films, thin film devices 1. Introduction Nanowire NbN superconducting single-photon detectors (SSPD) [1] is a novel type of single-photon detector successfully competing with InGaAs single-photon avalanche photodiodes at standard telecom wavelength of 1550 nm. Due to efficient coupling to single-mode optical fibres (above 90%) and high performance (detection efficiency 15% at the fibre input at 10 dark counts per second at 1550 nm wavelength, up to 100 MHz counting rate and 40 ps timing jitter [2]) SSPD was successfully used in applications ranging from single-photon source characterization [3, 4, 5] to quantum cryptography [6, 7]. Moreover, recently demonstrated successful coupling of the SSPD with optical waveguides makes these detectors a device of chose for optical signal processing on a single chip [8, 9]. Although it was demonstrated that the SSPD had a single-photon response up to wavelength of 5.6 μm its application at wavelengths longer than 1.5 μm was hampered by a dramatical drop of quantum efficiency η with the wavelength increase. Recently we demonstrated that reduction of the strip width to 56 nm improves η at wavelengths longer than 2 μm [10]. In this paper we further develop this approach and present the results of the SSPD characterization at 10.6 μm wavelength Published by Elsevier B.V. Selection and/or peer-review under responsibility of the Guest Editors. Open access under CC BY-NC-ND license. doi: /j.phpro

2 A. Korneev et al. / Physics Procedia 36 ( 2012 ) Fig. 1. SEM image of the parallel-wire SSPD with 40-nm-wide superconducting NbN strip. 2. Device topology and fabrication The initial trigger of the SSPD photoresponse is the formation of local region with suppressed superconductivity called hotspot in the place where a photon was absorbed. The size of the hotspot depends on the photon energy. It is worth noting that in a typical NbN nm wide strip made of 3 4 nm thick film the portion of the strip cross-section covered by the hotspot produced by a photon of 1 1.5μm wavelength is much smaller compared to the total strip cross-section. Thus the quantum efficiency of the detector can be improved if one reduces the strip width nevertheless maintaining its uniformity. We managed to improve our electron-beam lithography and reactive ion etching process so that we were capable to produce 40-nm-wide uniform strips. Meanwhile the reduction of the strip width imposes a certain restriction on the response pulse voltage. Indeed, the reduction of the strip width leads to the reduction of the critical current, and the bias current as well. And as the response voltage is roughly proportional to the bias current, its reduction results in the reduction of the voltage pulse making it hard to distinguish the pulse against the thermal noise background of the read-out electronics. One have several options here how to overcome this limitation, e.g. use HEMT amplifier operated in liquid helium and featuring reduced noise compared to the room temperature electronics, or one may try SQUID read-out. We applied a different approach. For response signal amplification we used an approach proposed in [11]: to connect the strips in parallel utilizing cascade switching mechanism. We managed to improve our electron-beam lithography process by using electron resist ZEP-520A-7 which was more robust in reactive ion etching and enabled higher resolution e-beam lithography compared to PMMA 950K, which we had used before. We used SEM JEOL 6380 upgraded to e-beam writer. We used 30 kv accelerating voltage with anode current ranging from 2 to 5 pa. As a developer we used ZED N50. During the e-beam process, we exposed the gaps between the superconducting strips. All strips were patterned having equal width. Due to such an improvement we were able to produce 40-nm-wide strips with 130 nm-wide gap between the strips. Our detectors had the total area of 7 μm 7 μm. Figure 1 presents an SEM image of our device. Due to large, area our detectors had 56 parallel strips. Thus the response mechanism of the detector should be similar to recently introduced arm-triggered regime [12] rather than simple cascade switching. 3. Experimental setup Recently we reported characterisation of parallel wire SSPD in wavelength range 1 μm 3.5 μm using a grating spectrometer as a light source [10]. A parallel-wire SSPD with 56-nm-wide strips exhibited an order

3 74 A. Korneev et al. / Physics Procedia 36 ( 2012 ) Fig. 2. Experimental set-up for parallel-wire SSPD characterization at 10.6 μm wavelength characterization. of magnitude better quantum efficiency at 3.5 μm wavelength compared to meander SSPD with 104-nmwide strip. Being encouraged by this result we moved forward to longer wavelength range and researched into the response of our new devices at 10.6 μm wavelength. The experimental setup is presented in Figure 2. As a light source we used CO 2 laser operated in cw (continuous wave) regime. We intentionally abandoned the usage of the grating spectrometer for long wavelength experiment and switched to the laser source. In a grating spectrometer together with the long wavelength (target wavelength) shorter wavelengths might also be present as a result of high orders of diffraction. Although these parasitic short wavelength are filtered out we were not absolutely sure that our spectrometer is completely free from such a defect. Usually these parasitic short wavelengths are not a problem as their power is several orders of magnitude smaller compared to the main wavelength of the beam. But as the quantum efficiency of the SSPD in visible and near infrared (at about 1 μm wavelength) is several several order of magnitudes higher compared to 10 μm wavelength this short wavelength light can considerably contribute to the count rate of the detector. Laser source operating at a single wavelength is free from such a defect. As CO 2 laser is a very powerful light source, its output beam was significantly attenuated. After that we placed a home-built variable attenuator enabling us to change the light power. Then we divided the beam by a beam-splitter. One part was sent to a Golay cell detector to monitor the attenuation introduced by our home-built attenuator. The other part was further attenuated and sent to the parallel-wire SSPD placed in an optical cryostat. The SSPD was maintained at 2 K temperature achieved by helium vapour evacuation. As the input window of the cryostat we used an filter transparent in 2 μm 25 μm wavelength range. To avoid latching we connected 3 Ω resistor in parallel. 4. Experimental results and discussion We measured detector dark count rate and light count rates as functions of detector bias current for various attenuations of the laser power. The results of these measurements are presented in Figure 3. The dark count rate (black squares in figure 3a) was measured by simply blocking the laser beam. One may notice that at low bias currents (below 700 μa) the dark counts rate deviates from the exponent typically observed (see e.g. [10]). We attribute such a deviation to the stray light and room temperature background. The light counts demonstrate a behaviour similar to the one previously observed at wavelengths 1 μm 3.5 μm [10]. Then we reploted count rates as a function of attenuation for bias currents of 550, 580, 600, 650, and 700 μa. The result is presented in Figure 3b. The probability of simultaneous absorption of m

4 A. Korneev et al. / Physics Procedia 36 ( 2012 ) Fig. 3. (a) Light count rates (open symbols) and dark counts rate (solid symbols) measured for different light attenuations; and (b) light count rate vs light attenuation demonstrate linear dependence as it is expected for single-photon response. photons within a certain time slot is given by the Poisson distribution: p en n m n!, (1) here n is the mean number of photons per time slot. For a cw source, as the time slot we take the lifetime of the resistive hot-spot in a strip of our parallel-wire SSPD. If the number of photons in the time slot n << 1, then equation 1 reduces to: p nm n!. (2) Thus one can resolve how many photons are responsible for the detector output voltage pulse. If the count rate is proportional to the mean number of photons n (i.e. power of the light source) then one photon is enough to trigger the detector. If the count rate is proportional to n 2 then the detector is triggered by photon pairs. If three photons are simultaneously required to trigger the detector then the count rate will be proportional to n 3. In Figure 3b we observe that although the points are scattered (due to inaccuracy in our home-built variable attenuator), in general they follow linear law. This fact suggests that single photon is enough to trigger the parallel-wire SSPD even at 10.6 μm wavelength. 5. Conclusion In conclusion, due to further improvement in the fabrication process we were able to produce SSPD with 40-nm-wide strips. We connected the strips in parallel to realize the amplification of the response voltage signal of the detector. Then we characterized the detector response at 10.6 μm wavelength with the cw laser source. We observed linear dependence of the count rate on the light attenuation as it is expected for single-photon response. 6. Acknowledgement This work was supported by Russian Federal Targeted Program Research and Development in Priority Areas of Russian S&T Development for the years References [1] A. Korneev, P. Kouminov, V. Matvienko, G. Chulkova, K. Smirnov, B. Voronov, G. Goltsman, M. Currie, W. Lo, K. Wilsher, J. Zhang, W. Slysz, A. Pearlman, A. Verevkin, R. Sobolewski, Sensitivity and gigahertz counting performance of NbN superconducting single-photon detectors, Applied Physics Letters 84 (26) (2004)

5 76 A. Korneev et al. / Physics Procedia 36 ( 2012 ) [2] [3] R. Hadfield, M. Stevens, R. Mirin, S. W. Nam, Single-photon source characterization with twin infrared-sensitive superconducting single-photon detectors, J. Appl. Phys. 101 (2007) [4] R. Hadfield, J. Habif, J. Schlafer, R. Schwall, S. W. Nam, Quantum key distribution at 1550 nm with twin superconducting single-photon detectors, Applied Physics Letters 89 (2006) [5] H. Takesue, S. W. Nam, Q. Zhang, R. Hadfield, T. Honjo, K. Tamaki, Y. Yamamoto, Quantum key distribution over a 40-db channel loss using superconducting single-photon detectors, Nature Photonics 1 (2007) [6] D. Stucki, C. Barreiro, S. Fasel, J.-D. Gautier, O. Gay, N. Gisin, R. Thew, Y. Thoma, P. Trinkler, F. Vannel, H. Zbinden, Continuous high speed coherent one-way quantum key distribution, Opt. Express 17 (16) (2009) [7] D. Stucki, N. Walenta, F. Vannel, R. T. Thew, N. Gisin, H. Zbinden, S. Gray, C. R. Towery, S. Ten, High rate, long-distance quantum key distribution over 250 km of ultra low loss fibres, New J. of Physics 11 (2009) [8] arxiv: v1 [physics.optics]. [9] arxiv: v1 [quant-ph]. [10] Y. Korneeva, I. Florya, A. Semenov, A. Korneev, G. Goltsman, New generation of nanowire nbn superconducting single-photon detector for mid-infrared, Applied Superconductivity, IEEE Transactions on 21 (2011) [11] M. Ejrnaes, R. Cristiano, O. Quaranta, S. Pagano, A. Gaggero, F. Mattioli, R. Leoni, B. Voronov, G. Goltsman, A cascade switching superconducting single photon detector, Appl. Phys. Lett. 91 (2007) [12] F. Marsili, F. Najaf, E. Dauler, F. Bellei, X. Hu, M. Csete, R. J. Molnar, K. K. Berggren, Single-photon detectors based on ultranarrow superconducting nanowires, Nano Lett. 11(5) (2011)