NIST Widens Superconducting Nanowire Single-Photon Detectors to 0.1 Millimeter and Cuts Dark Counts by Ten Orders of Magnitude, Published in Optica on August 19, 2026 Under DARPA's SynQuaNon Program
Quentir Defense Monitor
Evidence-based insights for quantum defense and security. Published by Quentir Systems LLC · September 17, 2026.
A NIST team in Boulder has shown that a superconducting single-photon detector can be made a thousand times wider than the usual design and still operate at the limit its material allows, with the dark count rate falling by ten orders of magnitude once a pair of current-carrying rails beside the wire is switched on. The paper by Kristen M. Parzuchowski, Eli Mueller, Adam N. McCaughan and fourteen colleagues appeared in Optica on August 19, 2026, was announced by NIST on August 24, and acknowledges DARPA's Synthetic Quantum Nanostructures program among its funders. For a defense reader the result is a receiver story. Photon-starved lidar, free-space laser links and low-light imaging all depend on how large, how quiet and how forgiving their detector is, and this work moves all three figures at once.
Superconducting nanowire single-photon detectors are the workhorse photon counters of quantum optics. A thin strip of superconductor carries a bias current just under the value at which it would stop superconducting. A single absorbed photon heats a small spot, the current is diverted around it, and a voltage pulse marks the arrival. NIST devices of this kind already register about 98 percent of incoming photons at telecom wavelengths. The catch has always been the width. Current in a superconducting strip crowds toward the edges over a characteristic distance called the Pearl length, which for the tungsten silicide films in this paper is about 600 micrometers. Any wire wider than a small fraction of that runs its edges hot and its center cold, so defects at the edges trigger false counts and the device has to be biased well below its potential. The practical answer for two decades has been a 100 nanometer wire wound into a meander, patterned by electron-beam lithography, sensitive to the polarization of the light, and hard to couple to anything except a single-mode fiber.
What the NIST team changed: niobium rails beside a tungsten silicide strip, and the numbers the Optica paper reports
The paper's device is a strip of tungsten silicide about 3 nanometers thick, in widths from 1 micrometer up to 100 micrometers, flanked on both sides by niobium rails about 50 nanometers thick and 4 micrometers wide. The rails carry their own current in the same direction as the detector, and the magnetic field they produce pushes the detector's current away from the edges and back toward the middle. With the rail current tuned to its optimum, the switching current of the 50 and 100 micrometer devices rose by 40 to 44 percent, the dark count rate fell by more than ten orders of magnitude, and the timing jitter dropped by roughly 30 percent to about 35 picoseconds. The arXiv version of the paper, first posted in January 2026 and revised through July, gives the 100 micrometer device as running at about a thousand dark counts per second with the rails off and at a rate too low to measure once they were on.
Two further results matter more than the headline width. First, the group showed near-unity internal detection efficiency at a wavelength of 4 micrometers on a 20 micrometer wide device, a factor of 20 wider than any mid-infrared photon counter previously reported at that efficiency. Mid-infrared photons carry about a quarter of the energy of telecom photons and make a smaller hot spot, which until now forced designers into ever narrower wires. Second, a wide straight strip has no meander, so it absorbs light of any polarization equally, and it presents a collecting area that free-space optics can fill without a fiber. The measurements were made at 0.9 kelvin, with the mid-infrared runs at about 260 millikelvin, and the paper notes that the rail current added roughly 120 millikelvin of heating through the cabling used. The NIST announcement quotes group leader Marty Stevens that it is unclear whether wide devices will reach the 98 percent system efficiency of their nanoscale relatives, and that more testing is needed.
The authors thank NIST, the University of Colorado, the National Academies and the DARPA Defense Sciences Office Synthetic Quantum Nanostructures program, which lists single-photon detector arrays for sensing, imaging and communication as one of its three technical thrusts. The coauthors span NIST, the two Colorado campuses, the Jet Propulsion Laboratory, the University of Glasgow, MIT and Old Dominion University. The journal version carries the DOI 10.1364/OPTICA.599984.
Quantum pillar: sensing (quantum radar and lidar). Use posture: dual-use. Technology readiness: TRL 4 of 9. Fabricated detectors were assembled and characterized under laboratory conditions at sub-kelvin temperatures, so the detector itself is validated in the laboratory, while any ranging, imaging or communications receiver built around it has yet to be demonstrated and would start lower on the ladder.
Why a wide, quiet, polarization-blind photon counter matters for lidar, free-space optical links and low-light imaging, and who gains
The capability reading starts with the dark count rate. In any photon-counting sensor, the noise floor sets the range. A single-photon lidar illuminates a scene with a weak laser and counts the handful of photons that come back, and how far it can see, or how little laser power it can use, is decided by how many false counts the detector produces in each timing bin. Ten orders of magnitude on that figure is the difference between a detector that has to be gated tightly around an expected return and one that can sit open and wait for a photon from a target whose range is unknown. Returns from a rough target arrive scattered across a patch of the focal plane and in random polarization, and a large, polarization-insensitive strip collects them without the coupling losses that a fiber-fed meander imposes. The mid-infrared result opens the 3 to 5 micrometer atmospheric window, where thermal signatures sit and where haze and smoke scatter far less than at visible wavelengths, to photon counting on a strip wide enough for ordinary photolithography.
The second application is free-space optical communication, and here the defense case is already documented. NASA's Deep Space Optical Communications experiment closed a laser link from the Psyche spacecraft to the Hale Telescope at Palomar using a photon-counting camera built by MIT Lincoln Laboratory around a 64-pixel array of superconducting nanowire detectors, holding 267 megabits per second at 19 million miles in December 2023 and 25 megabits per second at 140 million miles in April 2024. A laser link is narrow, hard to intercept and hard to jam, and the receiver technology that reads it at interplanetary distance is the same technology this paper makes larger and quieter. A ground or ship terminal that can accept a wandering, depolarized beam onto a wide detector is a simpler terminal than one that must steer the beam into a fiber core.
The third is low-light imaging, the strand where Europe's ADEQUADE consortium handed its ministries a single-photon imaging roadmap in June, as Quentir read in its post on that project's close. An imager needs many pixels, and every pixel adds its own dark counts to the frame. A pixel design that is intrinsically quiet and can be made large lets an array trade pixel count for collecting area without drowning in noise.
Who gains follows from the uses. The same device serves a hospital measuring blood flow through tissue, an observatory looking for faint galaxies and a dark-matter search. For a force, the value is on both sides of the line the Monitor draws. A quieter, wider receiver lets one's own laser links and terrain-mapping lidar run at lower power and longer range, which is protective. It also lets a sensor find a platform whose owner relies on distance, darkness or obscurants to stay hidden, which is adversary-facing. The honest posture is dual-use.
What stands between a 0.9 kelvin bench result and a receiver a program office can buy
The measurements are single-device laboratory measurements at sub-kelvin temperatures. Fielded SNSPD systems today run in closed-cycle cryocoolers at 2 to 4 kelvin. The mid-infrared result at 260 millikelvin needs a sorption or dilution stage, and the paper's own note that the rails added about 120 millikelvin of heating through their cabling shows how tight the thermal budget is. Wider devices will need their own optical stacks, since the cavity and anti-reflection design that gives a narrow meander its 98 percent system efficiency has not been shown for a 100 micrometer strip. Every device also needs a second bias line for its rails, which is a wiring and heat-load cost that multiplies in an array. The maximum count rate of a large strip, set by its kinetic inductance, is the figure a lidar or communications designer will ask for next, and the released text does not settle it. Nothing in the paper is a ranging, imaging or communications demonstration. It is a detector characterization. The applications above are inferences the authors and DARPA themselves draw.
That distinction matters for how a buyer reads DARPA's role. The Synthetic Quantum Nanostructures program is a small basic-research vehicle, and what it bought here is a physics result. Quentir's reading of DARPA's optical-clock purchase of August 2026, where the agency paid for a production run rather than another discovery, is the comparison to keep in view. For wide photon counters the discovery has just happened, and the copies, the arrays, the cryogenic packaging and the qualification pathway are all still ahead.
The paper establishes that the width ceiling on superconducting photon counters was a current-distribution problem and that a pair of rails removes it, with a measured ten-order improvement in noise and a measured mid-infrared efficiency on a device 20 times wider than before. It does not establish system efficiency, array behavior, count rate or operation above one kelvin. A program that depends on photon-starved ranging, on laser links through a contested spectrum, or on imaging in the mid-infrared should put this work on its watch list. The next results to look for are a wide device with a measured system efficiency in a fiber-free optical stack, an array built on the rail design, and a demonstration inside a 2 to 4 kelvin cryocooler.
Sources
Primary source: Kristen M. Parzuchowski, Eli Mueller, Bakhrom G. Oripov, Benedikt Hampel, Ravin A. Chowdhury, Sahil R. Patel, Daniel Kuznesof, Emma K. Batson, Ryan Morgenstern, Robert H. Hadfield, Varun B. Verma, Matthew D. Shaw, Jason P. Allmaras, Martin J. Stevens, Alex Gurevich and Adam N. McCaughan, "Reaching the intrinsic performance limits of superconducting nanowire single-photon detectors up to 0.1 mm wide," Optica, published online August 19, 2026 (arXiv 2601.15971, v4 of July 23, 2026). Other material: NIST's August 24, 2026 news release; DARPA's Synthetic Quantum Nanostructures program page; NASA's Deep Space Optical Communications mission page.