A superconducting nanowire single-photon detector (SNSPD) scaled to 0.1 mm wide — roughly 1,000× the 100-nm linewidths standard in the field — has recorded a billion-fold drop in dark counts at the National Institute of Standards and Technology, according to a paper published online in Optica on Aug. 19, 2026.

The work, led by NIST postdoctoral researcher Kristen Parzuchowski with coauthors Eli Mueller and electronics engineer Marty Stevens, attacked a long-standing trade-off in photon counting. Narrow wires catch nearly every photon but limit the current the strip can carry; wider wires were believed to lack the energy density needed to register a single photon's disruption of superconductivity.

What changed in the device geometry?

Conventional SNSPDs use a 100-nm-wide superconducting wire patterned as a meander on a cryogenic substrate. A photon striking the wire deposits enough energy to flip a small region from superconducting to normal, producing a measurable voltage pulse. The best NIST devices register about 98% of incoming photons.

The team widened the active wire two orders of magnitude and added two parallel superconducting "rails" that carry current in the same direction as the central strip. The rails' magnetic field combines with the central wire's field to redistribute current uniformly across the width.

"The rails let us push the device closer to the transition between the superconducting and normal state," Mueller said. "If you're operating even closer to that transition, then you're still sensitive to the very low energy photons."

That operating regime is what delivered the billion-fold dark-count reduction. The new architecture is also polarization insensitive — the detector registers photons regardless of the orientation of the electric field.

What does the underlying physics buy you?

The hot-spot mechanism requires a photon to break superconductivity across the full wire width. At 100 nm, a near-infrared photon carries enough energy to do that easily. At 100 µm, the same photon's energy is a smaller fraction of the energy needed to drive a wider strip normal, which is why previous attempts stalled. Flattening the current profile lets more supercurrent flow through the wire than edge-defect-limited geometry allowed, regenerating a hot spot at any practical width.

"For years, researchers have tried to get closer to the optimum performance of these detectors, but it was never clear how far you could push it," Parzuchowski said. "Now we've shown that you can actually reach the intrinsic performance limit."

The group has not yet measured detection efficiency on the wide device. Stevens said the 98% benchmark set by narrow-wire SNSPDs has not been demonstrated at 0.1 mm and requires fresh measurements. The fabrication simplification is, however, decisive: the wide wire is patterned by standard lithography rather than sub-100-nm tooling that constrains production scale.

Where does the architecture fit?

Applications the NIST team names — diffuse correlation spectroscopy for blood-flow imaging and faint-light astronomy — share two constraints: low photon flux, and a need for large active areas that narrow-wire meanders cannot cover without optical stacking.

"Typically, everyone has worked to make smaller and smaller wires, which makes fabrication increasingly challenging," Parzuchowski noted. The wider device sidesteps that pinch.

The lead author added a candid reaction to the dark-count result. "It was very shocking," she recalled. "We were calling in some other people in our group into the room, telling them, 'Look at this, this is crazy!'"

What has to be measured next?

The next questions are metrological. Can a 0.1-mm wire match the 98% detection efficiency of a narrow-wire device? Can the rail geometry scale to centimeter-class arrays without reintroducing edge crowding? Stevens indicated the team believes the wire can go wider still; the paper claims a demonstrated upper limit, not a maximum.

Funding came in part from the Defense Advanced Research Projects Agency under the Synthetic Quantum Nanostructures (SynQuaNon) program. The DOI is 10.1364/OPTICA.599984.

For quantum networks, deep-space optical links, and biomedical instruments that count photons one at a time, the result reframes a fabrication bottleneck as a design parameter — one whose efficiency ceiling is still unmapped.