3D cavity-based graphene superconducting quantum circuits in two-qubit architectures

TL;DR

This work demonstrates graphene-based superconducting circuits integrated with 3D cavities to realize flux-tunable qubits and multi-qubit coupling architectures. By loading graphene transmon devices into cavities with different resonant frequencies, the authors access multiple coupling regimes, observe vacuum Rabi splitting, and reveal two-stage dispersive shifts in a two-qubit configuration. The results establish a path toward scalable, multi-qubit 3D transmon devices built from 2D materials and highlight the feasibility of joint readout and inter-qubit interactions in graphene-based circuits. Overall, the study showcases flexible qubit–cavity coupling and motivates further development of 2D-material–enabled 3D quantum processors.

Abstract

We construct a series of graphene-based superconducting quantum circuits and integrate them into 3D cavities. For a single-qubit device, we demonstrate flux-tunable qubit transition, with a measured $T_1$ $\approx$ 48 ns and a lower bound estimate of $T_2^\ast$ $\approx$ 17.63 ns. By coupling the device to cavities with different resonant frequencies, we access multiple qubit-cavity coupling regimes, enabling the observation of vacuum Rabi splitting and flux-dependent spectral linewidths. In a two-qubit device consisting of a SQUID and a single junction, power-dependent measurements reveal a two-stage dispersive shift. By flux-tuning the cavity frequency at different readout powers, we attribute the first shift to the fixed-qubit and the second to the SQUID-qubit, indicating successful coupling between the two circuits and a single cavity mode. Our study demonstrates the flexible coupling achievable between 2D-material-based superconducting circuits and 3D cavities, and paves the way toward constructing multi-qubit 3D transmon devices from 2D materials.

Figures (12)

• Figure 1: Optical micrograph of the two-qubit graphene-based superconducting quantum circuit coupled to a 3D copper cavity. (a) Image of qubit chip mounted in the 3D copper cavity. (b) Optical micrograph of device 2, consisting of a SQUID-based qubit (Q1) and a single-JJ qubit (Q2). The inset shows the hBN/graphene/hBN sandwich structure with NbTi edge contacts for the junctions shown in (c) and (d). (c) Optical micrograph of the SQUID circuit consisting of graphene and superconductor NbTi. The design parameters can be found in Figure \ref{['FigS1']}(g). (d) Optical micrograph of the single-JJ circuit. The design parameters can be found in Figure \ref{['FigS1']}(g). (e) Schematic of the two qubits coupled to a microwave resonator. Q1 is coupled to the resonator $via$ the capacitance $C_{g1}$, while Q2 is coupled $via$$C_{g2}$. Q1 is coupled to Q2 $via$$C_{g12}$.
• Figure 2: Flux- and power-dependent cavity response (S$_{21}$) of the single-qubit device (Q1 in device 1) in cooldown 2. (a) Flux modulation of the cavity frequency for device 1. Note that the upper part and the bottom part were measured separately in order to improve the visibility of cavity response. The arrows indicate the flux bias points where the power-dependence measurements in (b) and (c) were performed. The dashed rectangle denotes the flux and frequency ranges displayed in (d). (b) Power-dependence measurement of device 1 at the zero-flux point (sweet spot) labeled (b) in (a). (c) Same as (b), but measured at the other flux point, where $f_q$ attains its minimum, as labeled (c) in (a). (d) Zoom-in measurement of the blue rectangular region indicated in (a), showing a vacuum Rabi splitting when the qubit frequency aligns with the bare cavity frequency. Note that the data have been mean-subtracted and smoothed along the flux axis for clarity. (e) Frequency splitting between the hybridized qubit-cavity states, as a function of $f_q$, as extracted from (a).
• Figure 3: Two-tone qubit spectroscopy of device 1 in cooldown 2. (a) Flux-dependent cavity response with the qubit frequency superimposed. Note that the data have been mean-subtracted for clarity. The dashed line indicates the inferred qubit frequency $f_q \approx f_{q,max}\sqrt{|\cos(\Phi)|}$, with a maximum value $f_{q,\mathrm{max}} \approx 6.438$ GHz as mentioned in the main text. (b) Two-tone measurement of the qubit transition as a function of applied flux $\Phi$ for device 1. The readout power was fixed at -50 dBm. The measurement was performed with two different drive powers: 15 dBm for the lower part of the spectrum, where $f_q$ is far detuned from $f_{bare}$, and -13 dBm for the upper part, where $f_q$ is close to $f_{bare}$. The spectrum is strongly distorted when the drive frequency approaches the bare cavity frequency ($f_{\mathrm{bare}} = 6.0558$ GHz). To resolve the maximal qubit frequency, a zoom-in measurement of the region indicated by the white dashed rectangle is performed and shown in the inset.
• Figure 4: Power dependence, flux-tuning and qubit spectroscopy for device 1 in cooldown 1. (a) Power dependence measurements of device 1 performed at $\Phi$ = 0 (sweet spot), with a dispersive shift $\chi/2\pi \approx 6.15$ MHz. (b) Flux modulation of the cavity frequency measured with a readout power of -60 dBm. The white dashed line indicates the bare cavity frequency at 6.059 GHz. (c) Two-tone measurement of the qubit transition as a function of applied flux, measured with a readout power of -60 dBm and a drive power of 8 dBm, showing a maximal qubit frequency $f_{q,\mathrm{max}} \approx 8.068$ GHz. The dashed lines denote the flux range for the analysis shown in (d). (d) Correspondence between the spectral linewidth ($\delta_{FWHM}$) and the flux derivative of qubit frequency ($\left|df_q/d\Phi\right|$) extracted from the spectrum in (c).
• Figure 5: Device 2 loaded into different cavities over multiple cooldowns. The first, second, and fourth cooldowns were performed using reflection measurements, as shown in Fig. \ref{['FigS2']}, whereas the third cooldown employed transmission measurements, as described in Ref. Chiu2025. (a) Power dependence of device 2 in a 6 GHz cavity, measured at the sweet spot (bottom panel), together with the flux modulation of the cavity response measured using VNA powers labeled A (middle panel) and B (top panel) in the bottom panel. (b) Same as (a), but measured in a 6.8 GHz cavity. (c) Same as (a), but using a transmission measurement setup. (d) Same as (a).