Anisotropic spin-valley coupling in SiMOS and Si/SiGe quantum dots

This study characterizes the angular dependence of spin-orbit coupling in SiMOS and Si/SiGe quantum dots, revealing that while both systems exhibit similar magnetic field orientations for minimizing spin-valley coupling, SiMOS devices display an order of magnitude stronger coupling than Si/SiGe, providing critical insights for optimizing silicon spin qubit performance.

The Gist

Imagine you are trying to build a super-fast, super-precise computer using tiny particles called electrons. In this computer, the "bits" of information aren't just 0s and 1s; they are tiny magnets called spins. To make these magnets work, scientists often use Silicon, the same material found in your phone's processor, because it's clean and quiet.

However, there's a catch. When you trap these electrons in a tiny box (called a Quantum Dot) to make them behave like computer bits, the walls of the box start to "talk" to the magnets in a weird way. This talk is called Spin-Valley Coupling.

Here is a simple breakdown of what the scientists in this paper discovered:

1. The Problem: The "Noisy" Room

Think of an electron's spin as a spinning top. In a perfect world, it spins perfectly straight up or down. But in Silicon, the "floor" of the room (the interface where the electron sits) is a bit bumpy.

• The Spin-Orbit Effect: Because the floor is bumpy, the spinning top starts to wobble. It doesn't just spin; it starts to tilt and sway.
• The Valley: Silicon has a special landscape with two "valleys" (like two dips in a hill). Usually, the electron sits in the lower valley. But if the wobble gets too strong, the electron might accidentally jump to the higher valley.
• The Danger: If the electron jumps to the wrong valley, it loses its information (the "bit" flips or disappears). This is called relaxation, and it's bad for the computer.

2. The Experiment: Two Different Houses

The researchers tested two different types of "houses" (devices) to see how bumpy the floors were:

• House A (SiMOS): A silicon house sitting on top of glass (Silicon Dioxide).
• House B (Si/SiGe): A silicon house sandwiched between two layers of a different material (Silicon-Germanium).

They wanted to know: Which house has the bumpier floor that causes the spinning top to wobble more?

3. The Discovery: The "Wobble" Direction Matters

The scientists didn't just look at the floor; they looked at how the wobble changed when they tilted the whole house using a giant magnet. They found two surprising things:

A. House A is much "wobblier" than House B.
The electrons in the SiMOS device (House A) wobbled about 10 times more than those in the Si/SiGe device (House B).

• Analogy: Imagine House A is a trampoline with loose springs, and House B is a trampoline with tight springs. If you jump on both, the loose springs (SiMOS) make you bounce and spin wildly, while the tight springs (Si/SiGe) keep you relatively stable.

B. The "Sweet Spot" is the same for both.
Even though House A was much wobblier, the direction in which the wobble was least bad was the same for both houses.

• Analogy: Imagine you are trying to balance a broom on your hand. If you hold the broom handle pointing North, it wobbles a lot. If you point it East, it wobbles less. The scientists found that for both types of houses, pointing the magnetic field in a specific direction (like Northeast) made the wobble the smallest.

4. Why This Matters: The "Goldilocks" Strategy

This research gives engineers a map for building better quantum computers.

• The Trade-off: Sometimes, you want the wobble. If the wobble is just right, you can use it to spin the electron and flip the bit (like turning a switch) without needing extra wires. This is great for speed.
• The Danger: If the wobble is too strong or in the wrong direction, the electron falls into the "wrong valley" and loses its data.

The Solution:
The paper suggests that by carefully choosing which material to use (SiMOS vs. Si/SiGe) and which direction to point the magnet, engineers can:

1. Avoid the danger: Point the magnet in a direction where the wobble is minimal to keep the data safe.
2. Use the power: Point the magnet in a direction where the wobble is strong enough to help flip the bits quickly, but not so strong that it breaks the computer.

Summary

Think of this paper as a guide for driving a car on two different types of roads (SiMOS and Si/SiGe).

• The SiMOS road is very bumpy (strong wobble), which makes the car shake a lot.
• The Si/SiGe road is smoother (weak wobble).
• However, on both roads, there is a specific lane (magnetic field direction) where the bumps are the least noticeable.

The scientists figured out exactly how bumpy each road is and where the smoothest lanes are. This helps engineers decide which road to build their quantum computers on and how to steer them to avoid crashes (data loss) while still driving fast.

Technical Summary

1. Problem Statement

Silicon is a leading material for spin qubits due to its weak intrinsic spin-orbit coupling (SOC) and the ability to isotopically purify it to eliminate nuclear spin noise. However, when electrons are confined to quantum dots (QDs) at interfaces (Si/SiO₂ in SiMOS or Si/SiGe wells), broken crystal symmetry induces significant interfacial SOC. This creates two critical challenges for qubit operation:

• g-factor Variability: Interfacial SOC causes variations in the electron g-factor between neighboring dots. While this can be harnessed for singlet-triplet (ST) qubit control, it introduces errors in exchange-only qubits that rely on magnetic field uniformity.
• Spin-Valley Relaxation: SOC couples spin states to excited valley states. When the Zeeman energy ($E_Z$) becomes resonant with the valley splitting energy ($\Delta_{vs}$), "spin-valley hot spots" occur. This leads to rapid spin relaxation ($T_1$) and decoherence, posing a major threat to qubit fidelity.

The paper addresses the need to characterize the angular dependence and magnitude of these SOC effects (both intravalley g-factor differences and intervalley spin-valley coupling) across the two dominant silicon qubit platforms: SiMOS and Si/SiGe.

2. Methodology

The authors performed comparative measurements on two distinct devices:

• SiMOS Device: A double quantum dot fabricated at Sandia National Laboratories using isotopically enriched $^{28}$Si on a thermal SiO₂ interface.
• Si/SiGe Device: A triple quantum dot fabricated at Intel using an isotopically enriched Si quantum well sandwiched between SiGe barriers.

Experimental Protocol:

1. Qubit Encoding: Both devices utilized a Singlet-Triplet ($S/T_0$) qubit encoded in two electrons. The qubit was initialized in the $(4,0)$ charge state and separated into the $(3,1)$ state where exchange coupling ($J$) was negligible.
2. Free Induction Decay (FID): The system was allowed to evolve under the effective magnetic field gradient ($\Delta E_Z$) driven by g-factor differences and spin-valley coupling. The rotation frequency between $|S\rangle$ and $|T_0\rangle$ was measured via Pauli spin blockade readout.
3. Angular and Magnitude Sweeps:
   • Magnetic field magnitude ($B$) was ramped to identify resonances where $E_Z \approx \Delta_{vs}$ (spin-valley hot spots).
   • Magnetic field orientation ($\theta, \phi$) was varied relative to the crystallographic axes ([100], [010], [001]) to map the anisotropy of the SOC.
4. Modeling: A four-level Hamiltonian model was constructed to describe the system, incorporating:
   • Intravalley SOC (Rashba and Dresselhaus terms) causing g-factor differences.
   • Intervalley SOC (spin-valley coupling, $\Gamma$) mixing ground and excited valley states.
   • The model was fitted to the experimental FID data to extract physical parameters.

3. Key Contributions

• Direct Comparison of Platforms: The study provides the first direct, quantitative comparison of SOC physics between SiMOS and Si/SiGe devices using identical measurement protocols and modeling frameworks.
• Quantification of Anisotropy: The authors mapped the full 3D angular dependence of the spin-valley coupling strength, identifying specific crystallographic orientations that minimize or maximize these effects.
• Physical Model Validation: They validated a four-level Hamiltonian model that successfully decouples the contributions of g-factor differences and spin-valley coupling to the observed ST rotation frequencies.
• Operational Strategies: The work proposes specific magnetic field orientations to either mitigate decoherence (avoiding hot spots) or exploit SOC for qubit control.

4. Key Results

The fitting of the model to the data yielded the following critical findings (summarized in Table I of the paper):

• Spin-Valley Coupling Magnitude:

  • SiMOS exhibits spin-valley coupling ($\gamma$) roughly one order of magnitude larger than Si/SiGe.
  • SiMOS: $\gamma \approx 0.73 - 0.87$ $\mu$eV.
  • Si/SiGe: $\gamma \approx 0.05 - 0.06$ $\mu$eV.
  • Implication: SiMOS qubits are more susceptible to spin relaxation near valley resonances but may offer stronger drive mechanisms for coherent control.

• Valley Splitting ($\Delta_{vs}$):

  • SiMOS devices showed significantly larger valley splittings (83–180 $\mu$eV) compared to Si/SiGe (37–47 $\mu$eV).
  • This suggests that the strong confinement at the Si/SiO₂ interface in SiMOS pushes excited valley states to higher energies, potentially offering a larger safety margin against accidental resonance, despite the stronger coupling.

• Angular Dependence:

  • Both systems exhibit similar anisotropic behavior. The spin-valley coupling is maximized when the magnetic field is aligned along the [110] and [$\bar{1}\bar{1}$0] crystallographic axes.
  • The coupling is minimized (nodes) when the field is oriented near the [$\bar{1}$10] axis (in the plane) or along the growth axis [001].
  • The g-factor difference ($\Delta g$) also shows angular dependence, maximized at specific azimuthal angles, but the phase offset differs slightly between the two material systems.

• Magnetic Field Dependence of Valley Splitting:

  • In the SiMOS device, the position of the spin-valley hot spots shifted with magnetic field orientation. The authors attribute this to magnetic field-dependent valley splitting, likely caused by magnetic confinement altering the sampling of interface disorder.

• Linewidth and Decoherence:

  • Near the spin-valley hot spots, the linewidth of the ST rotation frequency broadens significantly (up to >5 MHz), indicating enhanced dephasing. This is attributed to increased sensitivity to charge noise when the system is near the valley resonance.

5. Significance and Implications

This work provides a crucial roadmap for optimizing silicon spin qubits:

1. Material Selection: Si/SiGe devices offer lower spin-valley coupling (reducing relaxation risk) but lower valley splitting (increasing the risk of accidental resonance). SiMOS offers higher valley splitting but stronger coupling, requiring careful field management.
2. Field Orientation Optimization:
   • To minimize g-factor differences (beneficial for exchange-only qubits), fields should be applied normal to the interface ($B \parallel [001]$).
   • To minimize spin-valley coupling (beneficial for $T_1$), fields should be oriented along the [$\bar{1}$10] direction.
   • To exploit SOC for driving, fields can be aligned along [110] where coupling is maximal.
3. Error Mitigation: Understanding the anisotropy allows operators to choose magnetic field directions that avoid "hot spots" where decoherence is maximized, thereby extending qubit coherence times.
4. Future Directions: The results highlight the need for atomistic modeling of interface disorder to explain the variability in coupling strengths and suggest that engineering the interface (e.g., strain, Ge concentration) can tune these parameters.

In conclusion, the paper demonstrates that while bulk silicon has weak SOC, the interface physics in quantum dots is complex and highly anisotropic. By characterizing these effects in both major silicon platforms, the authors provide the necessary data to engineer robust, high-fidelity silicon spin qubits.