Toward single-photon detection with superconducting niobium diselenide nanowires

TL;DR

This work demonstrates single-photon detection using few-layer NbSe$_2$ nanowires fully encapsulated with hBN, preserving superconductivity after nanofabrication and enabling on-chip photonic integration. The devices achieve high optical responsivity across 650–1550 nm, with a 1/e recovery time around 135 ns and system timing jitter near 1.1 ns, while showing a detectable, albeit low, system detection efficiency at near-critical bias. Attenuated pulsed measurements provide evidence of single-photon sensitivity, yielding up to 33% pulse-detection probability at 1 MHz and a linear count response with photon number. The results position NbSe$_2$-based SNSPDs as a viable, scalable platform for ultrathin, crystalline detectors in quantum photonics, with clear pathways for integration and performance optimization.

Abstract

We present superconducting nanowire single-photon detectors (SNSPDs) based on few-layer NbSe$_2$ fully encapsulated with hexagonal boron nitride (hBN), demonstrating single-photon sensitivity. Our fabrication process preserves the superconducting properties of NbSe$_2$ in nanowires, as confirmed by low-temperature transport measurements that show a critical temperature of $T_c \approx 6.5$ K, comparable to the reported values for unpatterned sheets, and it maintains a contact resistance of $\sim 50 \, Ω$ at $T = 4$ K. Meandered NbSe$_2$ nanowires exhibit a responsivity of up to $4.9 \times 10^4$ V/W over a spectral range of 650-1550 nm in a closed-cycle cryostat at 4 K, outperforming planar and short-wire devices. The devices achieve a $1/e$ recovery time of $τ= (135 \pm 36)$ ns, system timing jitter of $j_\text{sys} = (1103 \pm 7)$ ps, and detection efficiency of $\sim 0.01\%$ at $0.95I_c$, with a linear increase in detection probability confirming the single-photon operation. Furthermore, measurements under attenuated pulsed laser (1 MHz) indicate a success rate of up to $33\%$ in detecting individual optical pulses, establishing the platform as a promising candidate for developing efficient single-photon detectors.

Figures (4)

• Figure 1: Fabrication of NbSe$_2$ nanowires. (a) Illustration of a fabricated NbSe$_2$ superconducting nanowire with top and bottom hBN encapsulation; the inset shows the crystalline structure of the van der Waals materials. (b) Optical microscopy image of an encapsulated 120 nm-wide nanowire with a resist mask patterned with a meander geometry; the scale bar is 10 µ m. (c) Scanning electron microscopy image of the nanowire; the scale bar is 1 µ m. (d) Five-point moving-average magnetoresistance of an encapsulated 500 nm-wide straight wire as a function of temperature, with the inset showing the field dependence of the critical temperature.
• Figure 2: Electrical characterization of NbSe$_2$ structures in a closed-cycle vacuum cryostat ($T = 3.9$ K). (a-c) I-V curves of (a) an unpatterned NbSe$_2$ flake with variable thickness, (b) A patterned, resist-encapsulated NbSe$_2$ meander wire, and (c) A patterned, fully encapsulated NbSe$_2$ meander wire, with their critical currents $I_c$ and retrapping currents $I_r$. The contact and wiring resistance was substracted for clarity. (d) Ratio of retrapping current to critical current for two wide flakes (F1 and F2), two short-wire samples (W1 and W2), and two meander samples (M1 and M2), characterized at the base temperature. NE = no encapsulation; FE = full encapsulation; RE = resist encapsulation.
• Figure 3: Optical characterization of NbSe$_2$ photodetectors. (a) Illustration of the experimental setup to optically characterize the NbSe$_2$ photodetectors. The device under test (DUT), represented by a variable resistor and an inductor, is cooled by a cold plate in the closed-cycle cryostat and irradiated with a focused laser beam. A bias voltage $V$ is applied to the circuit, which is completed by a high-load resistor ($R_s = 100 \text{ k}\Omega$) and, when using an oscilloscope or a time-to-digital converter, a parallel resistor ($R_p = 50 \, \Omega$); the detector response is either measured at the voltage source (when using CW lasers) or at the oscilloscope or time-to-digital converter after an amplifier. (b) Comparison of the voltage response of four samples under laser irradiation, normalized by the amplitude as the devices saturate ($T = 5$ K). The meanders (M1 and M2) and the wire (W1) are irradiated with a 650 nm CW laser, and the flake (F1) is irradiated with a 1080 nm pulsed laser. The incident power is normalized by the irradiated area of the device. (c) Voltage response of sample M1 at three CW laser wavelengths ($T = 5$ K). Optical response of the unpatterned flake irradiated by a 1064 nm pulsed laser with an 8 kHz repetition rate. (d) Voltage responsivity of sample M1 against the bias current at 220 nW of incident power for three CW laser wavelengths ($T = 4$ K).
• Figure 4: Toward NbSe$_2$ SNSPDs. (a) Illustration of the detection of attenuated laser pulses on the encapsulated meander sample, resulting in voltage spikes. (b) A typical pulse measured with an oscilloscope. The recovery time, measured as the $1/e$ decay constant, is calculated by averaging 138 peaks. (c) Measurement of the timing jitter $j_\text{sys}$ of the entire system in the experimental setup. (d) Detected counts at the TDC against the number of photons (femtosecond, 1100 nm) in each pulse for three current bias values. The lines are obtained by fitting the points above the noise count level. The error bars show one standard deviation. (e) Detection efficiency and dark count rate against the normalized bias current (normalized by the critical current). (f) Detected laser pulses against the normalized bias current.