Fast and Continuous Detection of Single Microwave Photons via Photo-assisted Quasiparticle Tunneling to a Superconducting Island

Abstract

We demonstrate a single-photon detector operating in the microwave domain, based on photo-assisted quasiparticle tunneling events that poison a superconducting island. The detection relies on continuously monitoring the island's charge parity using microwave reflectometry. This scheme achieves 10% detection efficiency with sub 50 ns time resolution and short dead time 1 microsecond, for microwave photons at 10 GHz. The detector features three junctions connected to a superconducting island, which together carry out photoelectric conversion and charge readout. The enhanced light-matter coupling, crucial to photon to quasiparticle conversion, is provided by a granular aluminum-based high-impedance microwave resonator. The time resolved detection of itinerant microwave photon opens up new perspectives in quantum sensing, microwave quantum optics, readout of superconducting qubits and mesoscopic physics.

Figures (7)

• Figure 1: Detection principle and device. (a) Simplified circuit diagram of the detector. An incoming microwave photon at the detector input is absorbed via photo-assisted tunneling into a superconducting island, inducing a quasiparticle-poisoning event. The resulting change in charge parity is continuously monitored through microwave reflectometry at the readout port. Three dc bias voltages ($V_r$, $V_c$, $V_g$) tune the working point of the detector. (b) Optical micrograph of the device. Both converter and readout resonators are made of 700 wide, 20 thick granular aluminum wires with large kinetic inductance ($0.5nH/\Box$). The converter resonator is 73 long and the readout resonator is made of two sections, which are 48 and 172 long. (c) Energy diagram representation of the microwave photo-assisted tunneling process. (d) Scanning electron micrograph of the superconducting island and the three junctions JC,JR and JD. The gate appears in purple.
• Figure 2: Charge stability with voltages. (a) Dependence of one quadrature, $I$, of the reflected readout signal on gate voltage and readout bias at zero converter bias. The observed Coulomb diamonds are $2e$ periodic and delimit stable charge states with a well defined number of Cooper pairs. The crosshair indicates the detector operating bias point. (b) Dependence of the readout signal at zero readout bias on gate voltage and converter bias. Selected gate voltage line cuts are overlaid with $2e$ (blue) and $e$-periodic (yellow) sinusoidal curves, illustrating the transition from a $2e$ to $e$ periodic response. (c) Probability of the odd charge state as a function of converter bias $V_c$ in absence of photon irradiation. The inset shows the histogram of the readout signal at $V_c=385\uV$ integrated for 100 using the JPA setup. The measured quadrature is rescaled and offset such that the two maxima are located at 0 and 1. The three vertical lines indicate the thresholds used to extract the odd-state probability $p_1$ shown in the main plot. Each curve corresponds to the threshold lines indicated in the inset.
• Figure 3: Photo-assisted Charge Parity Jump. (a) Increase in the odd-charge-state probability, $p_1(N) - p_1(0)$, after sending a pulse containing $N$ photons at 10GHz to the detector input. The colored regions, each of width $h \nu/e$, indicate the voltage ranges where $P$-photon-assisted tunneling processes are allowed by energy conservation. The shaded area represents the $1\sigma$ statistical uncertainty. (b) Detection probability $p_1(N)-p_1(0)$ as a function of $N$ at $V_c=385\uV$ (vertical line in a). The data are obtained with the JPA. Inset: Frequency dependence of the detection probability measured at $N=0.8$; the Lorentzian fit yields a FWHM linewidth of 82MHz.
• Figure 4: Continuous photon detection (a) Time traces of the readout signal binned with a resolution of 48ns. The reconstructed charge parity identified by the Viterbi algorithm are shown as purple traces. An incoming pulse with 1.12 photon is applied at $t = 5µs$. (b) Averaged trace obtained from 4096 repetitions of the measurement in (a), showing the pulse arrival and subsequent recovery of the detector. (c) Probability of detecting an even to odd parity jump as a function of time near $t = 5µs$.
• Figure S1: Schematic of the measurement setup cabled in the dilution refrigerator.