Yachin Ivry, Jonathan J Surick, Maya Barzilay, Chung-Soo Kim, Faraz Najafi, Estelle Kalfon-Cohen, Andrew D Dane and Karl K Berggren
Detecting single photons at infrared wavelengths is central to some next-generation communication and sensing technologies and a key requirement of quantum-cryptography and computing. Superconducting nanowire single-photon detectors (SNSPDs) are capable of detecting photons at the near IR with: >90% system efficiency , timing jitter less than 30 ps, reset time of just a few nanoseconds and with lower dark-count rate than competing technologies. For example, the recent loop-hole free demonstration of violation of Bell’s inequality was made possible the availability of these detectors. Similarly, recent demonstrations of ultrafast space-based optical communications relied on these detectors. Future progress in the domain of quantum and classical communication and information processing thus requires continued reduction in jitter and reset time, and further increases in efficiency of this technology.
However, existing SNSPD materials have not been optimized to allow all of these advantages to occur simultaneously, e.g., there is a fundamental tradeoff between speed and efficiency—detectors with high efficiency tend to require larger areas, and thus have longer reset times and larger jitter. Optimization of this tradespace is hindered by the existence of a relatively limited number of superconducting materials that are available for use in SNSPDs, and the large number of detector parameters that have to be optimized, including: operating temperature, optical absorption at or across a variety of wavelengths, fabrication yield, material stability under thermal cycling, output signal amplitude, and resistance to dark counts. Tungsten silicide (WxSi1−x), niobium nitride (NbN), niobium titanium nitride (NbTiN), magnesium diboride, yttrium barium copper oxide, molybdenum germanium, molybdenum silicide, and niobium have all been used as superconducting nanowire detectors, although currently NbN, NbTiN, and WxSi1−x based SNSPDs are preferred, having demonstrated detection efficiencies approaching the quantum limit (observed as a saturating behavior of the detection efficiency as a function of bias current applied on these devices). Systems that include such devices have exhibited over 90% total system efficiency with dark count rates as low as <1 s−1 . However, such WxSi1−x-based detectors have a number of drawbacks, higher kinetic inductance and requiring lower operating temperatures (<1 K and can be as low as 120 mK) . This situation invites us to consider modifying materials, to find routes to somehow engineer superior properties in a single device.
Previous attempts to externally tune superconducting properties of sensitive photodetectors include mainly the utilization of proximity effects between a superconductor and a normal metal. That is, when a normal metal is introduced in close proximity to a superconductor, the normal electrons can diffuse into the superconductor, suppressing the superconducting transition temperature TC. This TC-tuning mechanism allowed e.g. in transition edge sensor, deterministic tuning of device properties, such as the detected photon wavelength (we should note that in principle, a counter effect where superconducting electrons are diffused into the normal metal is also possible, but is less common in superconducting-based photodetection technologies). An obvious approach to tuning the materials of NbN would then be to deposit a thin layer of normal layer on top of it. However, given that we know tungsten silicide (WSi) has advantageous detector properties that are somewhat complementary to those of NbN, we decided to attempt to proximitize the NbN with WSi (i.e. another superconductor), rather than with a normal metal.