Content uploaded by Roman Sobolewski
Author content
All content in this area was uploaded by Roman Sobolewski
Content may be subject to copyright.
Ultimate sensitivity of superconducting single-photon detectors in
the visible to infrared range
A. Verevkin1, J. Zhang, A. Pearlman, W. Slysz, and Roman Sobolewski
University of Rochester, Rochester, NY 14627-0231, USA
A. Korneev, P. Kouminov, O. Okunev, G. Chulkova, and G. Gol’tsman
Moscow State Pedagogical University, Moscow 119435, Russia
ABSTRACT
We present our quantum efficiency (QE) and noise equivalent power (NEP) measurements of the meander-
type ultrathin NbN superconducting single-photon detector in the visible to infrared radiation range. The
nanostructured devices with 3.5-nm film thickness demonstrate QE up to ~10% at 1.3 – 1.55 µm
wavelength, and up to 20% in the entire visible range. The detectors are sensitive to infrared radiation with
the wavelengths down to ~10 µm. NEP of about 2 × 10-18 W/Hz1/2 was obtained at 1.3 µm wavelength.
Such high sensitivity together with GHz-range counting speed, make NbN photon counters very promising
for efficient, ultrafast quantum communications and another applications. We discuss the origin of dark
counts in our devices and their ultimate sensitivity in terms of the resistive fluctuations in our
superconducting nanostructured devices.
INTRODUCTION
Superconducting single-photon detectors (SSPDs) [1] based on ultrathin, submicron-width NbN
superconducting meander lines have already found several practical applications [2-4], based on their
picosecond-range response time, low jitter, and low dark counts. The SSPD ability to efficiently count
single photons has been explained within a phenomenological hot-electron photoresponse model [1, 5].
This model describes formation and subsequent growth of a resistive hotspot in a very narrow, current-
biased superconducting stripe upon the single-photon absorption event [6], [7]. The hotspot formation is
followed by supercurrent redistribution, leading to a transient voltage response signal [8]. The NbN-based
SSPDs are able to efficiently detect single photons in the wavelength range from ultra-violet to near
infrared (IR) [4], [9] at a GHz-range counting speed [10]. They work at the 4-K temperature level, which
allows a long-term, no-interruption work, when using modern closed-cycle refrigerators. Their fiber-optical
coupling makes easy to perform room temperature photon-counting experiments.
The aim of this presentation is to demonstrate the performance of our new SSPDs, fabricated from
3.5-nm-thick NbN films. We found a significant (up to two orders of magnitude) increase of quantum
efficiency (QE) in such detectors, as compared with the 10-nm-thick structures [4]. Our new 3.5-nm
devices exhibit QE up to ~10% at near IR and above 15% in the visible range. We associate such an
improvement of QE with the large increase of a hotspot size in thinner NbN films. We also present here the
sensitivity performance of our detectors. Knowing the dark counting rate and its bias current dependence
we were able to determine a noise equivalent power (NEP), which is widely regarded to be the best
measure of the detector sensitivity. For our best devices the NEP reaches 2 × 10-18 W/Hz1/2 at 1.3 µm
wavelength.
EXPERIMENTAL DETAILS
The devices used in our experiments were 10 × 10-µm2-area size NbN superconducting meander-
type structures with the film thickness d = 3.5 nm, the nominal stripe width w ~ 200 nm, and the gap
between the meander stripes of about 300 nm (filling factor ~0.4). The details of our fabrication process are
Contact information for A. Verevkin- Email: verevkin@ece.rochester.edu
described in [11]. Critical current density of devices varied within 5-8 MA/cm2 range at 4.2 K and
superconducting transition temperature Tc was about 10 K.
The devices under study were wire-bonded to a 50-Ω microwave stripe line and mounted on a
cold base plate maintained at 4.2 K in vacuum inside either a liquid-helium dewar or a two-stage
cryocooler. Voltage pulses generated by the SSPD under illumination were amplified by 1-2 GHz
bandwidth, room-temperature amplifier and fed to a photon counter. Radiation from different laser sources
was delivered inside the cryostat through free space, or using a standard multimode 50-µm-core-diameter
fiber, and focused by collimator on the device plane inside the cryostat. In NEP measurements the
incoming 1.3-µm radiation from a luminescence diode was filtered inside cryostat with a cold (4.2 K) 0.9-
1.6 µm bandpass filter to block long-wavelength thermal background as well as any parasitic visible light.
EXPERIMENTAL RESULTS
Our latest 3.5-nm-thick devices demonstrate significantly larger QE, as compared to the previous,
10-nm-thick devices [3]. The improvement is about two orders of magnitude in the visible range, and even
larger for IR photons. Figure 1 presents QE versus wavelength for a 10 × 10-µm2 area, 3.5-nm-thick NbN
SSPD. We believe that in the visible range our latest SSPDs have already reached the limiting intrinsic QE,
corresponding to counting of all photons incident upon the device’s superconducting stripe. As the result,
QE tends to be independent of radiation wavelength in the visible range. In addition, the entire spectral
dependence shown in Fig. 1 is significantly flatter than the dependence characteristic for the 10-nm-thick
devices [4]. The reason for such a significant improvement of QE in the thinner SSPDs is quite obvious –
the normal resistance hotspot area in those devices is significantly larger, as it should be roughly inversely
proportional to the film thickness [6].
Fig. 1. Spectral dependence of QE for a typical 10x10-µm2 area, 3.5-nm-thick NbN meander detector. Temperature is
4.2 K with the current bias is equal to 0.9Ic.
The other parameter, besides QE, responsible for ultimate sensitivity of single-photon detectors is
the dark counting rate R. R, together with QE, determines the NEP of single-photon detectors in the
following manner:
2
h
NEP R
QE
υ
=. (1)
The R(I) dependence, shown in Fig 2 (right vertical axis), demonstrates the activation law R =
a×exp(b×I/Ic) in the whole biasing range used in our experiments (0.98 > I/Ic > 0.85), where a and b are
constants, and Ic is the detector critical current. The activation-type behavior of R(I) can be observed within
five orders of magnitudes range and is independent of the device size or the actual superconductor stripe
width. Left vertical axis in Fig. 2 corresponds to QE dependence on the normalized current bias. As in the
case of R, we observe an exponential decrease of QE with the I/Ic ratio decrease, but at least for bias
currents close to Ic, the QE drop is relatively slow.
Figure 3 presents NEP versus I/Ic, obtained using Eq. (1) and the experimental results on both QE
and R shown in Fig. 2 within the 0.98 > I/Ic > 0.85 range (closed dots). We note that NEP of about 2 × 10-18
W/Hz1/2 has been achieved for I/Ic = 0.87. This is the best sensitivity in the near-IR range, ever reported for
a single-photon detector operating at 4 K temperature level. As we can see in Fig. 3, the NEP rapidly
decreases with the bias decrease, since close to Ic, R is a much stronger function of I/Ic then QE.
Extrapolating the R data from Fig. 2 and extending the NEP plot into low bias using experimental QE
points from Fig. 2, we can reach NEP below 10-20 W/Hz1/2.
Fig. 2. QE and dark counting rate R versus the normalized bias current, measured for a 3.5-nm-thick, 10x10-µm2 area
SSPD at 1.3-µm wavelength. Solid line illustrates the activation-type law of the R dependence on the detector bias.
For low I/Ic values, the NEP saturates, due to the very fast drop of the QE curve at low current bias (Fig.
2). The physical meaning of this saturation phenomenon is quite simple – the detector becomes insensitive
to incident radiation and the QE slope follows the R slope, as the supercurrent redistribution is too small to
produce a voltage response at such low biases.
Fig. 3. NEP versus the normalized bias for a 3.5-nm-thick, 10x10-µm2 area SSPD at 1.3-µm wavelength. Closed circles
correspond to experimental data presented in Fig. 2, while open circles represent the extrapolated result based on the
measured QE and the R (see Fig. 2) interpolated to the low bias currents. Dashed line is a guide to the eye.
DISCUSSION
Dark counts and their rate R play the basic role for the ultimate sensitivity of any single-photon
detector, as we can see from Eq. (1). The origin of dark counts in our NbN SSPDs in absence of any
incident photon flux or significant thermal background fluctuations represents an interesting physical
problem. We believe that the observed dark counts are mainly due to resistive (or, equivalently,
superconducting) fluctuations in our NbN superconducting stripes. Such fluctuations were observed in one-
dimensional (1-D) superconducting stripes at low temperatures and have been discussed in terms of
quantum phase slip centers (PSCs) [12-14]. We have already pointed out to the role of such fluctuations in
the context of the experimentally measured, exponential spectral dependence of QE [4].
The 1-D picture is not fully adequate for our devices. The thickness d of the NbN superconducting
stripe is indeed close to the BCS coherence length
ξ
, but the stripe width w is significantly larger than
ξ
. On
the other hand, the 1-D approach seems to be productive to improve our understanding of the problem and
for development of a quantitative model of dark counts in our NbN SSPDs. A significant extension of the
1-D phase slips formation process [15] towards the case of wider stripes (w >>
ξ
) was obtained at low
temperatures at the presence of bias currents close to Ic [16,17]. It was shown in [16] that the essential
features of the PSC model are preserved, when compared with the 1-D case, including the quasiparticle
diffusion length, and the drop of the pair chemical potential over the
ξ
distance. A general, theoretical
investigation of the problem in the 2-D case was proposed in [18], and it was recently used to study of PSC
formation in Nb and YBCO bridges [19]. The main conclusion made in [19] was that the PSCs could occur
in superconducting stripes more frequently than it is generally admitted by modern theories for uniform
clean superconductors at temperatures well below Tc. Authors in [19] suggested that this latter observation
could support the idea proposed in [18], namely that the PSCs are formed from microscopic normal-state
hotspots existing in granular films.
We strongly believe that the above case applies to our NbN films. Detailed, atomic-force
microscope and electron microscope studies did not show any evidence of nm-scale defects in our NbN
films. However, our structures are characterized by a negative resistivity temperature coefficient within the
20-300 K temperature range, what is often associated with the existence of granular microstructure in thin
NbN films [20]. Consequently, in agreement with the concept proposed in [18],[19], the origin dark counts
in our structures is very likely to be associated with granularity of our NbN films. More detailed study on
this subject will be presented elsewhere.
This work was funded by the US Air Force Office for Scientific Research grant F49620-01-1-0463
(Rochester) and by the RFBR grant 02-02-16774 (Moscow). Additional support was provided by the
NPTest, San Jose, CA.
REFERENCES
1. G. N. Gol’tsman, O. Okunev, G. Chulkova, A. Lipatov, A. Semenov, K. Smirnov, B. Voronov, A.
Dzardanov, C. Williams, and R. Sobolewski, Appl. Phys. Lett. 79, 705 (2001).
2. S. Somani, S. Kasapi, K. Wilsher, W. Lo, R. Sobolewski and G. Gol’tsman, J. Vac. Sci. Technol.
B 19, 2766 (2001).
3. R. Sobolewski, A. Verevkin, G. N. Gol’tsman, A. Lipatov, and K. Wilsher, IEEE Trans. Appl.
Supercon. (June 2003).
4. A. Verevkin, J. Zhang, R. Sobolewski, G. Chulkova, A. Korneev, A. Lipatov, O. Okunev, A.
Semenov, and G. N. Gol’tsman, Appl. Phys. Lett. 80, 4687 (2002).
5. A. D. Semenov, G. N. Gol’tsman, and A. A. Korneev, Physica C 351, 349 (2001).
6. A. M. Kadin and M. W. Johnson, Appl. Phys. Lett. 69, 3938 (1996).
7. K.S. Il’in, I. I. Milostnaya, A. A. Verevkin, G. N. Gol’tsman, E. M. Gershenzon, and R.
Sobolewski, Appl. Phys. Lett. 73, 3938 (1998).
8. J. Zhang, W. Slysz, A. Verevkin, R. Sobolewski, O. Okunev, and G. N. Gol’tsman, Phys. Rev. B
67, 132508 (2003).
9. A. Lipatov, O. Okunev, K. Smirnov, G. Chulkova, A. Korneev, P. Kouminov, G. Gol'tsman, J.
Zhang, W. Slysz, A. Verevkin, and R. Sobolewski, Supercon. Sci. and Tech. 15, 1689 (2002).
10. J. Zhang, W. Slysz, A. Verevkin, O. Okunev, G. Chulkova, A. Korneev, A. Lipatov,
G. N. Gol’tsman, and R. Sobolewski, IEEE Trans. Appl. Supercon. (June 2003).
11. G. N. Gol’tsman, K. Smirnov, P. Kouminov, B. Voronov, N. Kaurova, V. Drakinsky J. Zhang,
A. Verevkin, and R. Sobolewski, IEEE Trans. Appl. Supercon. (June 2003).
12. C. N. Lau, N. Markovic, M. Bockrath, A. Bezryadin, and M. Tinkham, Phys. Rev. Lett. 87,
217003 (2001).
13. M. Tinkham and C.N. Lau, Appl.Phys. Lett 80, 2946 (2002).
14. A. Bezryadin, C. N. Lau, and M. Tinkham, Nature 404, 971 (2000).
15. W.J. Skocpol, M.R. Beasley, and M. Tinkham, J. Low Temp. Phys. 16, 145 (1974).
16. V.G. Volotskaya, I.M. Dmitrenko, L.E. Musienko, and A.G. Sivakov, Sov. J. Low Temp. Phys. 7,
188 (1981); Sov. J. Low Temp. Phys. 10, 179 (1984).
17. A. Garzarella and C.J. Martoff, J. Appl. Phys. 79, 2426 (1996).
18. A. Weber and L. Kramer, J. Low. Temp. Phys. 84, 289 (1991).
19. J.-P. Maneval, F. Boyer, K. Harrabi, and F.-R. Ladan, J. of Superconductivity 14, 347 (2001).
20. J.-H. Tyan, and J.T. Lue, Journal of Appl. Physics 75, 325 (1994).
