On-Demand Correlated Errors in Superconducting Qubits from a Particle Accelerator

Abstract

Ionizing radiation is a known source of correlated errors in superconducting quantum processors, inhibiting the functionality of quantum error correction surface codes. High-energy photons and charged particles deposit pair-breaking energy into these systems leading to excess quasiparticles near Josephson junctions that increase qubit decoherence. Previous investigations of this problem have relied on ambient, stochastic sources of ionizing radiation or alternative methods of quasiparticle generation. Here, we present a facility that couples an electron linear accelerator (linac) to a dilution refrigerator to study ionizing radiation in quantum systems. A single linac electron closely mimics the energy deposition characteristics of a typical cosmic-ray muon, and we demonstrate the facility's usefulness with a multi-qubit superconducting transmon chip. Characteristic radiation-induced relaxation errors are quickly and easily collected with the speed and timing information of the linac. Additionally, we present qubit excitation and detuning errors that can be difficult to detect without the on-demand source of ionizing radiation. These error signatures are shown to be dependent on the junction placement and surrounding superconducting gaps.

Figures (13)

• Figure 1: Experimental Design. (a) Schematic of the CLIQUE facility. The electron beam (red dashed line) exits the linac, redirects through the bending magnet, and travels into the adjoining room where it strikes the dilution refrigerator. The linac trigger (blue dotted line) is supplied to both the linac and the quantum control hardware for event tagging. A scintillation detector (star) is positioned just off-axis from the incident beam to continuously monitor the electron fluence. The scale bar (lower right) is 1 m. (b) Scintillator response to electron fluence tuning. The probability distribution of energy deposition in the scintillator is collected for two linac gun grid voltage pulses. Vertical gray dashed lines mark the expected energy deposition for 1, 2, and 3 electrons. (c) MCNP simulation comparison of the energy deposition distribution for linac electrons and typical cosmic-ray muons in a 350 µ m silicon chip. The most-probable energy deposition, 100 k eV, is marked with a vertical gray dashed line. (d) Josephson junction design, indicating the superconducting gaps of the layers and their orientations relative to the capacitor island or ground plane for the two layouts (green vs. red labeled qubits). The junction interface is shown in pink, where QPs (unpaired electrons) can tunnel across and cause qubit errors. (e) Design schematic of the qubit chip, with labeled qubits and junction locations highlighted in pink. (f) Relaxation detection measurement. All qubits are prepared in the $\ket{1}$ state with a 1 µ s detection delay before measurement. The lighter blue curve is a single correlated error event coincident with the linac trigger and the darker blue points are the average of all 831 detected linac events across $\sim$7.4 min of data.
• Figure 1: CLIQUE Facility. (a) Image of the linear accelerator (left) and bending magnet (right). (b) Image of the quantum hall, with dilution refrigerator and gas handling system mounted on a movable rail framework. (c) Image of the wall penetration (illuminated with blue light) and modified mixing chamber stage shield (windows removed). (d) Image of the qubit device packaging (gray, center) viewed from the accelerator hall through the wall penetration.
• Figure 2: Design-dependent relaxation and excitation errors. (a) Error signature for $\ket{1}$ state preparation with 1 µ s detection delay averaged across all detected linac events for low-gap-island qubits [green in Fig. \ref{['fig:intro']}(e)]. Narrow lines are exponential fits ($A e^{-t/\tau} + c$) for each qubit. (b) Data collected coincidentally with that of (a) and analyzed similarly, for high-gap-island qubits [red in Fig. \ref{['fig:intro']}(e)]. (c) Error signature for $\ket{1}$ state prep (blue) and $\ket{0}$ state prep (orange) with 1 µ s detection delay across the largest 20% of detected linac events for q22 (low-gap-island). The darker narrow curves are fits using the rate equation dynamics described in the main text. The inset shows the same data and fits in a shorter time window following the linac trigger. (d) Data collected coincidentally with that of (c) and analyzed similarly for q23 (high-gap-island). The inset shows the q23 results for a 12 µ s detection delay using the same fit parameters.
• Figure 2: Scintillator Fluence Calibration. Measured electron fluence at the location of the silicon die versus the electron fluence outside the dilution refrigerator. A linear fit is overlaid on the data and shows a ratio of 0.177 between the inside and outside fluences.
• Figure 3: Detuning and dephasing errors. (a) Time slices following the linac trigger of the Ramsey-like detection scheme for q22. Solid curves are the averaged data across 542 detected events and dashed lines are decaying sinusoidal fits ($A e^{-t/T_\phi} \sin{(\omega t_{detect} + \phi_0)} + c$). (b) Extracted detuning (blue circles) and dephasing rates (orange triangles) for q22. The plotted detuning excludes the intentional offset of the Ramsey experiment. Solid lines are exponential fits to the data (dashed to extend back to the trigger time). (c) Comparison of detuning, dephasing, and relaxation error recovery times for all qubits. Error bars in (b)and (c) are standard deviations from the fits.