Fast and Sensitive Readout of a Semiconductor Quantum Dot Using an In-Situ Microwave Resonator with Enhanced Gate Lever Arm

TL;DR

The paper tackles the need for fast, high-fidelity readout in Si/SiGe double quantum dots by integrating an on-chip Nb coplanar-stripline resonator directly with the gate electrodes and optimizing the gate lever arm $\alpha_g$. Through coupled Schrödinger–Poisson simulations and careful device design, the authors maximize capacitive coupling to achieve a dispersive readout with $g_{\mathrm{eff}}=\alpha_g g_0$ and demonstrate unity SNR at $\tau_{min}=34.54$ ns, corresponding to a $14.48$ MHz detection bandwidth and a charge sensitivity of $\delta q\approx1.86\times10^{-4}\,e/\sqrt{\mathrm{Hz}}$. Analysis of the baseband $I/Q$ PSD reveals a $1/f$-type charge-noise floor below ~10 kHz, linking performance to material and fabrication noise sources. The work establishes lever-arm engineering as a scalable, fabrication-friendly strategy to enable fast, high-fidelity readout suitable for real-time feedback and potential integration with quantum error correction in semiconductor qubit architectures, while outlining practical paths to mitigate charge noise and further enhance coupling.

Abstract

We report an experimental study of a Si/SiGe double quantum dot (DQD) directly coupled to a niobium superconducting coplanar stripline (CPS) microwave resonator. This hybrid architecture enables high-bandwidth dispersive readout suitable for real-time feedback and error-correction protocols. Fast and sensitive readout is achieved primarily by optimizing the DQD gate lever arm, guided by MaSQE quantum dot simulations, which enhances the dispersive signal without requiring high-impedance resonators. We demonstrate a signal-to-noise ratio (SNR) of unity with an integration time of 34.54 nanoseconds, corresponding to a system bandwidth of 14.48 MHz and a charge sensitivity of 0.000186 e per square root hertz. Analysis of the voltage power spectral density (PSD) of the in-phase (I) and quadrature (Q) baseband signals characterizes the system's readout noise, with the PSD's dependence on integration time providing insight into distinct physical regimes.

Figures (11)

• Figure 1: Illustration of experimental setup. Experiments were conducted in an Oxford Instruments Triton dilution fridge with base temperature of 60 mK. The microwave generator signal is split to device and IQ-mixer. Cold stage attenuation is used to avoid thermal noise, which could lead to qubit dephasing and thermal broadening. The single port systems is made into a two port measurement setup by using a circulator to send the input microwave signal to the device and the output reflected signal can be amplified and compared to the input; isolating reflections. We measure the real ( I ) and imaginary ( Q ) part of the reflection coefficient, $S_{11}$ and apply DC offsets to the plunger gates using a quantum machines OPX. A 4K stage low noise amplifier (LNA), by low noise factory is used to amplify the signal from the devices, this is critical to attain a signal. The labels on the device zoom-in are L for left, R for right, P for plunger, and B for barrier.
• Figure 2: SEM of device used for the experiments. The design is an accumulation mode device with overlapping gates.
• Figure 3: Schematic of the heterostructure and device design.
• Figure 4: (a) Measured reflection coefficient magnitude and phase as a function of the left plunger gate voltage as well as mathematical fits described by Equations \ref{['manuscript:eq:voigt']} and \ref{['manuscript:eq:tanphase']}. (b) Same as panel (a) but as a function of the right plunger gate voltage. (c) Loaded quality factor and resonant frequency based on fits for each plunger gate value sweep.
• Figure 5: Experimentally obtained stability diagram when the device was tuned into a DQD. The slopes of the charging lines relate to the lever arm matrix elements as can be seen in Supplementary Materials \ref{['manuscript:sup:lever arms']}. We find that the plunger gates have an effect of $27\%$ and the cross capacitance is $5\%$, this was in good agreement with the simulation results.