arXiv:2610.31998
Demonstrating RFSoC-based Control of a Fully Connected Exchange-Only Qubit
Abigail L. Wessels, M. Garrett Bimstefer, Joseph D. Broz, Alen Senanian, Andrew E. Oriani
Abstract
Semiconductor triple-quantum dots (TQDs) are a promising platform for scalable quantum information processing, but the community has so far lacked access to affordable and flexible high-speed control electronics. SpinQICK, an open-source spin qubit-control package built on the Xilinx Radio Frequency System-on-Chip (RFSoC), directly addresses this gap by leveraging the high sampling-rate signal generation and advanced digital signal processing (DSP) capabilities of the Quantum Instrumentation Control Kit (QICK) to enable precise time-domain voltage control and readout of semiconductor spin qubits. In this paper, we use spinQICK to demonstrate coherent control, state preparation, and readout of a fully connected triangular exchange-only device. We also introduce and experimentally verify a novel dynamical decoupling (DD) sequence enabled by the triangular TQD and leverage its unique decoupling properties to characterize electrical noise. These results establish spinQICK as a low-cost, flexible platform for semiconductor spin qubit control and provide a pathway for hardware optimization through qubit-level noise diagnostics.
AI Summary
The paper tests whether an open-source RFSoC control stack can support preparation, coherent exchange control, and readout of a triangular, fully connected exchange-only spin qubit, and whether dynamical decoupling can diagnose noise introduced by the control electronics. The spinQICK implementation provides 14 baseband DAC channels operating at approximately $6.88$ GSPS, two readout-stimulus channels, FPGA-based readout demodulation, and programmable crosstalk compensation on 11 channels. Experiments demonstrate exchange oscillations on all three couplings, using predistorted $10$ ns pulses, and Pauli-blockade readout with typical signal-to-noise ratios of 5–6. The principal sequence contribution is $[NZM]^2$, termed NZM1, which uses the third exchange axis to decouple quasistatic hyperfine noise while retaining an initialization-dependent quasistatic exchange-noise channel. Measurements comparing NZM1 with the existing $[NZ]^3$ sequence distinguish magnetic noise, electrical noise, and calibration drift. Stochastic simulations using experimentally inferred noise spectra reproduce several initialization-dependent decay envelopes, but not oscillations attributed to temporally correlated drift. The protected NZ1y lifetime is approximately $60\,\mu\mathrm{s}$, versus approximately $600\,\mu\mathrm{s}$ in a prior comparable experiment. A fitted electrical white-noise floor of approximately $40\,\mathrm{nV}/\sqrt{\mathrm{Hz}}$ is attributed primarily to room-temperature amplification. NZM1 oscillations indicate calibration drift fitted at $65.7$ Hz/s. The demonstrated setup therefore remains limited by amplifier noise, long-timescale instability, and pulse distortion requiring $40$ ns interpulse idles. The precise drift source is unresolved; simulations assume shot-to-shot uncorrelated noise and omit excited orbital and valley states.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.