🔬 mesoscale physics

Device characterization of Si $/$ SiGe double quantum dots using exchange oscillations in Earth's magnetic field

This paper demonstrates that intrinsic magnetic-field gradients from residual nuclear spins in Si/SiGe double quantum dots can be utilized to perform exchange oscillations and characterize device parameters at zero applied magnetic field, offering a simple, high-throughput diagnostic tool for hybrid semiconductor-superconductor qubits without the need for micromagnets or complex calibration.

Original authors: Holly G. Stemp, Harry Hanlim Kang, Chih Hwan Yang, Gabriel D. Cutter, Frederike Brockmeyer, Patrick J. Strohbeen, Max Hays, Jeffrey A. Grover, William D. Oliver

Published 2026-09-01

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Holly G. Stemp, Harry Hanlim Kang, Chih Hwan Yang, Gabriel D. Cutter, Frederike Brockmeyer, Patrick J. Strohbeen, Max Hays, Jeffrey A. Grover, William D. Oliver

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the quest to build a quantum computer, scientists are looking for tiny switches that can hold information in a state of superposition, existing in multiple possibilities at once. One promising path involves trapping individual electrons in tiny wells of silicon, using their spin—a fundamental property that makes them act like microscopic magnets—as the carrier of information. To make these electron spins talk to each other and perform calculations, researchers rely on a force called exchange interaction, which links the spins of neighboring electrons. However, for these systems to work reliably, the environment must be incredibly quiet. Any stray magnetic noise can scramble the information, causing the computer to lose its memory. Traditionally, controlling these spins has required applying strong external magnetic fields or using tiny, custom-made magnets on the chip. But for some advanced designs that mix silicon with superconducting materials, even a small magnetic field can ruin the delicate superconducting components, creating a difficult trade-off between control and stability.

A team of researchers has now found a way to bypass this dilemma by using the Earth's own magnetic field and the natural, random magnetic noise inside the silicon itself to drive the system. In a study published in September 2026, scientists from the Massachusetts Institute of Technology and collaborators demonstrated a method to characterize and control these silicon-based quantum bits without applying any external magnetic field. They utilized a double quantum dot device, which is essentially a pair of tiny traps holding two electrons, fabricated from silicon and silicon-germanium. Instead of fighting against the natural magnetic variations caused by the atomic nuclei within the silicon, the researchers turned these variations into a tool. By carefully timing electrical pulses, they were able to make the two electron spins oscillate between different states, using the tiny, intrinsic magnetic differences between the two dots to drive the motion.

The researchers discovered that by using a specific sequence of electrical pulses, they could extend the time the quantum information remained coherent, or intact, from a brief 1.17 microseconds to a much more usable 74.8 microseconds. This was achieved by applying up to seventy refocusing pulses that effectively canceled out the random noise, a technique known as dynamical decoupling. This extension of coherence time is crucial because it gives the quantum computer more time to perform calculations before the information degrades. The team also used these oscillations to map out the "idle" behavior of the device, measuring how much the two electrons interacted with each other when they were supposed to be doing nothing. They found that this unwanted interaction, or residual exchange, varied across the chip, being stronger near the edges of the operating area and weaker in the center. By measuring the phase of the oscillations, they could detect these tiny interactions, which occur at frequencies in the tens of kilohertz, and map them out to find the best spots for operation.

Furthermore, the study revealed how the system behaves when the interaction between the electrons is turned up high. The researchers observed that the rate at which the quantum state decays changes depending on the strength of the interaction. They found a broad region where the decay rate increased significantly when the interaction frequency was between 20 and 40 megahertz. This suggests that at these specific strengths, the system becomes more susceptible to noise or leaks into other states, a detail that is vital for engineers trying to design stable quantum processors. The ability to detect these subtle effects without needing external magnets or complex calibration procedures offers a simpler, more direct way to test and tune these devices. This approach is particularly valuable for hybrid systems that combine silicon with superconductors, where magnetic fields are strictly limited. By proving that intrinsic magnetic gradients can be harnessed for precise control and characterization, the work provides a practical diagnostic tool that could accelerate the development of large-scale quantum computers built from silicon.

Problem
Exchange-based semiconductor qubits, such as singlet-triplet and exchange-only qubits, are promising candidates for scalable quantum processors due to their small footprint and compatibility with standard fabrication. However, their reliable operation requires precise characterization of environmental noise, residual idle interactions, and exchange-dependent decay. Conventional characterization methods often rely on multi-axis control calibration or the deliberate engineering of magnetic-field gradients using integrated micromagnets or dynamic nuclear polarization (DNP). These requirements present significant challenges, particularly for hybrid semiconductor-superconductor architectures where external magnetic fields can degrade superconducting components. Furthermore, DNP is difficult to implement in isotopically enriched materials where the nuclear spin ensemble is small. There is a need for a simpler, zero-applied-field diagnostic tool that can characterize these devices without micromagnets, DNP, or prior multi-axis calibration.

Methodology
The authors utilize a Si/SiGe double quantum dot (DQD) device fabricated using single-layer etch-defined gate electrodes (SLEDGE). The device is operated at zero externally applied magnetic field. Instead of an external gradient, the authors exploit the intrinsic magnetic-field gradient ( $\Delta B_z$ ) generated by the hyperfine coupling between confined electrons and the residual $^{29}$ Si nuclei (approx. 800 ppm) and $^{73}$ Ge nuclei in the barriers.

The core technique involves driving exchange oscillations between the two quantum dots using this intrinsic gradient. The pulse sequence is analogous to a Hahn-echo experiment:

Initialization: A singlet state is initialized in the (0,2) charge configuration.
Dephasing: The system is moved to the (1,1) charge configuration for a free-evolution time $\tau$ , where the exchange interaction $J$ is suppressed ( $J \ll g\mu_B \Delta B_z$ ). During this time, the spin state precesses about the intrinsic $\Delta B_z$ axis.
Refocusing: A voltage pulse is applied to the barrier gate (X5) and plunger gates (P5, P6) to rapidly increase the exchange interaction ( $J \gg g\mu_B \Delta B_z$ ) for a duration $\tau_p$ . This pulse rotates the state about the exchange axis.
Rephasing: The system returns to the (1,1) configuration for a second free-evolution period of time $\tau$ .
Readout: The singlet return probability is measured via Pauli spin blockade.

By sweeping the amplitude or duration of the exchange pulse, the authors tune the rotation angle to be odd or even multiples of $\pi$ . Odd multiples ( $\pi, 3\pi, \dots$ ) refocus the dephasing accumulated from the nuclear gradient, while even multiples do not. The authors employ Carr-Purcell-Meiboom-Gill (CPMG) sequences with up to $N=70$ refocusing pulses to extend coherence. They also analyze the phase of the oscillations to quantify residual exchange during idle periods and measure singlet decay rates as a function of exchange frequency.

Key Contributions and Results

Extended Coherence: Using CPMG sequences with up to 70 refocusing pulses, the authors extended the singlet coherence time from a free-induction decay time of $T_2^* = 1.17 \pm 0.02$ $\mu$ s to a CPMG coherence time of $T_2^{\text{CPMG}} = 74.8 \pm 1.8$ $\mu$ s.
Residual Exchange Mapping: The phase of the exchange oscillations was used as a sensitive probe to resolve residual exchange coupling in the tens-of-kilohertz regime during idle periods in the (1,1) charge cell. By comparing sequences with different numbers of refocusing pulses ( $M=1$ to $4$), the authors mapped the residual exchange across the (1,1) charge cell, finding values ranging from $30 \pm 5$ kHz to $64 \pm 8$ kHz depending on the operating point.
Decay Rate Characterization: Measurements of singlet-return decay under constant exchange interaction revealed a broad enhancement in the decay rate between exchange frequencies of approximately 20 and 40 MHz. The authors suggest this may be due to enhanced relaxation/leakage via state hybridization or increased longitudinal dephasing noise at the exchange splitting.
Intrinsic Gradient Utilization: The study demonstrated that the intrinsic nuclear spin gradient is sufficient to drive exchange oscillations and perform high-fidelity refocusing without external magnetic fields or micromagnets. The measured nuclear gradient width ( $\sigma_{Bz}/2\pi \approx 140 \pm 6$ kHz) was consistent with expectations for the isotopically enriched material.

Significance
The paper establishes intrinsic-gradient exchange oscillations as a simple, relevant diagnostic tool for exchange-only and related semiconductor qubit encodings. The method is particularly valuable for hybrid semiconductor-superconductor systems where magnetic fields are detrimental. By using the dephasing mechanism itself to generate a measurable singlet-return response, the technique avoids reliance on specific microscopic origins of dephasing, making it broadly applicable to other zero-field dephasing mechanisms refocusable by exchange pulses. The authors claim this approach enables rapid, high-throughput device characterization of coherence, idle interactions, and decay processes without the need for complex magnetic field engineering or multi-axis calibration.

More like this