Reaching the intrinsic performance limits of superconducting strip photon detectors up to 0.1 mm wide

Abstract

Superconducting nanowire single-photon detectors (SNSPDs) have emerged as the highest performing photon-counting detectors, making them a critical technology in quantum photonics and photon-starved optical sensing. However, the performance of SNSPDs is limited not by the intrinsic properties of the superconducting film, but by edge-induced current crowding. Despite extensive materials optimization and increasingly demanding fabrication strategies aimed at mitigating this edge-limited behavior, the device edges continue to limit the maximum device operating current, thereby degrading key performance metrics. Here, we demonstrate for the first time in situ tuning of a detector from an edge-limited to a bulk-limited regime, allowing the device to reach its intrinsic performance limit. Our approach is based on current-biased superconducting "rails" placed on either side of the detector to suppress current crowding at the edges. We show that activation of the rails reduces the dark count rate by nine orders of magnitude and extends the photon detection plateau at 1550 nm by more than 40%. These results are demonstrated on detectors up to 0.1 mm wide, establishing an entirely new class of ultra-wide strip detectors that we call superconducting strip photon detectors (SSPD). Moreover, the ability to suppress edge current crowding using the rails provides a pathway toward SSPDs with strip widths extending into the mm-scale. Such devices will enable large-area, high efficiency SSPD arrays with infrared sensitivity and open new opportunities in applications ranging from biomedical imaging to deep space optical communication.

Figures (9)

• Figure 1: (a) Illustration of SSPD (gray) integrated with adjacent rails (gold). The rails are displaced slightly from the edge of the strip. Simulated magnetic field lines are shown to demonstrate how the rails modify the self-induced magnetic field of the SSPD. (b) SEM image of 50 $\mu$m-wide WSi strip with 4 $\mu$m-wide Nb rails. The inset shows a zoomed-in image of the constriction. (c) Normalized current density as a function of position $x$ along the wire width $w$ simulated for a bare strip without rails (dark blue) and for a strip tuned by a series of increasing rail currents (purple to pink). For these simulations, the strip width is $w = 0.08 \Lambda$ and the rails are displaced from the edges of the strip horizontally and vertically by $w/200$. (d) Measured dark count rate as a function of $I_{\mathrm{s}}$ on a 50 $\mu$m-wide strip for a series of increasing $I_{\mathrm{r}}$ values from 0 mA (dark blue) to 11.8 mA (yellow). Dark count rates onset at higher $I_{\mathrm{s}}$ as $I_{\mathrm{r}}$ is increased. (e) Measured 1550 nm photon count rate (left vertical axis, data points) and dark count rate (right vertical log-scale axis, lines) as a function of $I_{\mathrm{s}}$ for rails off (dark blue) and rails on (pink) for a 50 $\mu$m-wide strip. Gold covers were placed over the tapers of the SSPD to prevent detection events outside the constriction, as the measurements were performed under flood illumination. The plateau width is $\approx 320\, \mu$A with rails turned off and extends to $\approx 450\, \mu$A with rails turned on. The start of the plateau was set as the value of $I_{\mathrm{s}}$ at which the count rate exceeds $10.7\times10^3$ s$^{-1}$. The end of the plateau was set as the value at which the dark count rate exceeds 100 s$^{-1}$.
• Figure 2: (a) Measured flood-illuminated 1550 nm photon count rate (left vertical axis, data points) or dark count rate (right vertical log-scale axis, lines) as a function of $I_{\mathrm{s}}$ for rails off (dark blue) or rails on (pink) for a 100 $\mu$m-wide strip. The plateau width is $\approx 540\, \mu$A with rails turned off and extends to $\approx 770\, \mu$A with rails turned on. The start of the plateau was set as the value of $I_{\mathrm{s}}$ at which the count rate exceeds 6.9$\times10^3$ s$^{-1}$. The end of the plateau was set as the value at which the dark count rate exceeds 100 s$^{-1}$. (b) Measured flood-illuminated 4 $\mu$m photon count rate (left vertical axis, data points) and background count rate (right vertical log-scale axis, lines) on a 20 $\mu$m-wide strip as a function of $I_{\mathrm{s}}$ for rails off (dark blue) and rails on (pink). At low $I_{\mathrm{s}}$ the background counts are dominated by blackbody background photons entering through the free-space coupled cryostat, then, as $I_{\mathrm{s}}$ is increased, the exponential dark counts takeover as indicated by the dashed gray line. Turning on the rails shifts the dark counts to higher $I_{\mathrm{s}}$ enabling measurements at higher $I_{\mathrm{s}}$. In this region, a rollover of counts is observed, indicating the start of a plateau and near unity IDE.
• Figure 3: Measured flood-illuminated 1550 nm photon count rate (pink data points) and dark count rate (lines) as a function of $I_{\mathrm{s}}$ for rails off (dark blue) and rails on (pink) on a 10 $\mu$m-wide strip with low $I_{\mathrm{sw}}/I_{\mathrm{d}}$ ratio. With the rails off, the dark counts begin to dominate at such a low $I_{\mathrm{s}}$ that photon counts cannot be measured. With the rails on, a huge shift of dark counts to higher $I_{\mathrm{s}}$ is observed, and photon counts are measured with near unity IDE. The slope on the plateau is consistent with photon detection events in the taper of the device.
• Figure 4: (a) Jitter histograms for a 20 $\mu$m-wide strip under 1550 nm flood illumination for six equally-spaced $I_{\mathrm{r}}$ values ranging from $I_{\mathrm{r}}=0$ to $I_{\mathrm{r}}=8.8$ mA. The long tail in the distribution for delay times greater than $\sim$50 ps is likely due to detection events in the taper as these measurements were performed under flood illumination. The jitter histogram narrows as $I_{\mathrm{r}}$ is increased. (b) FWHM jitter as a function of $I_{\mathrm{r}}$ for two 5 $\mu$m (yellow), two 10 $\mu$m (pink) and three 20 $\mu$m (dark blue) wide strips. As $I_{\mathrm{r}}$ is increased, the jitter decreases until it reaches a minimum.
• Figure 5: Measured $I_{\mathrm{sw}}/I_{\mathrm{d}}$ as a function of $I_{\mathrm{r}}$ for strip widths of 20 $\mu$m (yellow), 50 $\mu$m (pink) and 100 $\mu$m (dark blue). As $I_{\mathrm{r}}$ is increased, $I_{\mathrm{sw}}/I_{\mathrm{d}}$ also increases until it reaches $I_{\mathrm{r}}^*$. Change ($\Delta$) in $I_{\mathrm{sw}}/I_{\mathrm{d}}$ as a function of $I_{\mathrm{r}}$ is shown in the inset for the same devices. As wire width is increased, larger increases in $I_{\mathrm{sw}}/I_{\mathrm{d}}$ are measured.