Engineering Ge profiles in Si/SiGe heterostructures for increased valley splitting

arXiv:2505.22295 · cond-mat.mes-hall · Submitted 2025-05-28 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.

Kai: I'm Kai, and with me are Mira and Lev, guest researcher.

Mira: Today's paper: "Engineering Ge profiles in Si/SiGe heterostructures for increased valley splitting".

Kai: Electron spin qubits in Si/SiGe quantum wells are limited by small and variable conduction band valley energy separations,

Mira: First, who's behind it and why it matters.

Paper summary: Mira: Thinking about the broader impact of this paper, the connection they established between the quantum Hall regime and simulated quantum dot behavior is quite compelling for condensed matter theory.

Kai: I agree, Mira; it provides a concrete link between macroscopic transport measurements in a 2DEG and microscopic physics within quantum dots, which helps ground our theoretical models <ref:2505.22295#pg1>.

Lev: For error correction research, this means we have a physical mechanism to target the energy splitting directly via material engineering rather than relying solely on external gate voltages or complex pulse sequences.

Mira: If we can reliably engineer that linear relationship between E QD v and E v, it gives us a powerful predictive tool for designing future qubit architectures in these Si/SiGe platforms.

Kai: The practical implication is that we can start designing specific material growth protocols aimed at achieving that factor of two enhancement in valley splitting while keeping the disorder manageable, as shown by structures B1 through B3.

Lev: From a hardware perspective, knowing the trade-off parameters—the interface width versus mobility loss—is essential for choosing which structure is right for our immediate experimental setup.

Mira: The work essentially provides a roadmap on how to tune material parameters to control fundamental quantum properties like valley splitting in solid-state systems.

Kai: So, this paper lays out a clear direction for experimentalists and theorists working with Si/SiGe spin qubits by suggesting profile engineering as a viable route to improved valley splitting.

Conclusion: Kai: So, we've been looking at how they actually built these structures and what they measured in those quantum Hall experiments, and now we need to look at the big picture with this paper's title and authors.

Mira: I think it’s important to understand that the core idea revolves around systematically changing the germanium concentration profiles within these Si/SiGe quantum wells to manipulate valley splitting directly.

Lev: From an error correction standpoint, if we can tune that splitting via material growth, it means we have a physical knob to adjust the energy scales involved in our qubit operation without needing massive external voltage sweeps.

Kai: Right, and the authors of this paper are showing exactly how they engineered these profiles—specifically mentioning those "increasingly thin quantum wells with intentionally diffused interfaces"—to get that tunability.

Mira: The implications for condensed matter theory are significant because they’ve established a concrete mechanism where interface morphology directly dictates the valley splitting energy in a two-dimensional electron gas.

Lev: If the correlation between the simulated quantum dot splitting and the 2DEG splitting holds up under real hardware conditions, it gives us a much more reliable way to model how these materials will behave as we scale up device sizes <ref:2505.22295#pg1>.

Kai: Exactly; this moves us closer to designing hardware where we can predict whether increasing disorder or thinness will help or hurt our qubit performance before we even start the fabrication.

Mira: It really hammers home that breaking translation symmetry through geometry is a viable path for controlling these fundamental electronic properties in solid-state systems.

Lev: This suggests that achieving higher valley splitting might actually be achievable with manageable disorder, which is a key hurdle for running robust quantum information processing.

Kai: So, the authors are essentially showing us the physical blueprint for how to build those specific heterostructures to achieve that enhanced splitting while keeping mobility respectable.

Mira: The real question now is whether this linear relationship they found between simulated and measured values can be perfectly replicated in a device with actual experimental disorder present at scale.

QuTech and Kavli Institute of Nanoscience, Delft University of Technology · University of Wisconsin-Madison · Catalan Institute of Nanoscience and Nanotechnology (ICN2) · CSIC and BIST · ICREA

cond-mat.mes-hall

Submitted: 2025-05-28

Updated: 2025-05-28

Journal ref: Nano Letters, 25, 12892-12898 (2025)

DOI: 10.1021/acs.nanolett.5c02848

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 79/100

The gist: Electron spin qubits in Si/SiGe quantum wells are limited by small and variable conduction band valley energy separations, which this work addresses by engineering germanium concentration profiles to

Key concepts

Valley Splitting
This refers to the energy difference between different valleys an electron can occupy within its conduction band. In Si/SiGe systems, this splitting is crucial for controlling spin states in semiconductor devices. The study focused on increasing this energy gap by engineering the material structure.
Germanium Concentration Profiles
The researchers precisely controlled where germanium atoms were placed within the quantum well structure. By making the interfaces intentionally diffused and thin, they could tune how much an electron's wave function overlaps with these Ge atoms, directly influencing valley splitting.
Quantum Hall Regime
This is a specific physical state of electrons in a 2DEG under a strong magnetic field where energy levels form discrete Landau levels. The study used measurements in this regime to probe the valley splitting energy gap ($\Delta_1$), which is then related to disorder and other gaps.
Trade-off between Splitting and Disorder
The paper found that increasing valley splitting often comes with higher electrical disorder, as seen by decreased mobility. However, the study identified a 'beneficial trade-off' where a specific structure provided an excellent balance: high enough splitting for qubit applications while maintaining respectable electron mobility.

Terminology

Summary

Electron spin qubits in Si/SiGe quantum wells are limited by small and variable conduction band valley energy separations, which this work addresses by engineering germanium concentration profiles to enhance valley splitting while maintaining low disorder. The core finding is that growing increasingly thin quantum wells with broad interfaces allows for a tunable enhancement of valley splitting, even though this correlates with increased electrical disorder in the two-dimensional electron gas (2DEG). This study provides a first insight into the connection between valley splitting in the quantum Hall regime and in quantum dots by showing a linear relationship between simulated quantum dot valley splitting energy and 2DEG valley splitting.

Engineering Ge Concentration Profiles

The researchers engineered germanium concentration profiles of 28Si/28SiGe heterostructures to increase the overlap of the electron wave function with Ge atoms in a tunable way. This was achieved by growing increasingly thin quantum wells with intentionally diffused interfaces. The study characterized these profiles using atomic-resolution scanning transmission electron microscopy (STEM) combined with secondary ion mass spectroscopy (SIMS) data. The resulting structures, B1–B3, featured interfaces that were approximately 2.4 times wider than the sharp-interface control heterostructure A. This engineering allowed for a tunable increase in valley splitting by adjusting the quantum well growth time and interface width.

Characterization of Disorder and Mobility

The disorder properties of the 2DEG were assessed using classical transport measurements, including Hall-bar-shaped heterostructure field effect transistors (HFETs) and magnetotransport measurements at 70 mK in a dilution refrigerator. The study found that while higher valley splitting correlates with increased electrical disorder, a beneficial trade-off between enhanced valley splitting and low disorder is achievable. Specifically, the maximum mobility decreased as the quantum well became increasingly thinner (from 3.8(4) × 105 cm2/Vs in B1 to 0.58 × 105 cm2/Vs in B3), which was compatible with the presence of Ge throughout the thin quantum well. However, similar low values of percolation density were found across all heterostructures, suggesting that alloy disorder only weakly affects the scattering rate at low density.

Quantum Hall Regime Measurements

Valley splitting was probed using activation energy measurements in the quantum Hall regime. The researchers focused on the first valley-split energy gap (∆1) at filling factor ν = 1. They extracted mobility gaps, Zeeman split gaps, and Landau gaps as a function of magnetic field B. A striking linear relationship was observed between these gaps and the disorder-induced Landau level broadening (Γ). This led to the estimation that the valley splitting energy is given by Ev = ∆1 + Γ. Furthermore, they determined that in heterostructures B1–B3, valley splitting increased linearly with B across the investigated range.

Connection to Quantum Dots

Simulations based on the experimental Ge concentration profiles revealed a linear relationship between simulated quantum dot valley splitting energy (E QD v) and the measured two-dimensional electron gas (2DEG) valley splitting energy (Ev). This correlation was quantified by calculating a dimensionless quantity, denoted as η1, which quantifies the overlap of the electron wave function with Ge atoms. The finding that Ev ∝ η1 mirrors theoretical predictions for average valley splitting in alloy disorder-dominated quantum dots. Consequently, simulations showed a linear relationship between E QD v and Ev at a specific orbital energy (1.88 meV), providing a predictive model for the enhancement of valley splitting in quantum dots supported by these heterostructures.

Conclusion and Trade-off

The study concludes that heterostructures B1–B3 could support on average increased valley splitting in quantum dots, as proxied by the measured valley splitting in the quantum Hall regime. They demonstrated that a quantum well with "much broader interfaces (≃3.6 nm) and similar width (≃ 7.8 nm) offers an excellent trade-off, featuring a 1.8× valley splitting increase, whilst still having respectable mobility (>2×105 cm2/Vs) and low percolation density (< 6×1010 cm−2)." This suggests that increasing valley splitting, which requires breaking translation symmetry, comes at the expense of a more disordered potential landscape.

The gist: Growing increasingly thin quantum wells with broad interfaces allows for a tunable enhancement of valley splitting while maintaining low disorder. The study provides a first insight into the connection between valley splitting in the quantum Hall regime and in quantum dots by showing a linear relationship between simulated quantum dot valley splitting energy and 2DEG valley splitting.

How it works

  1. Growing increasingly thin quantum wells with intentionally diffused interfaces: This technique was used to control the overlap of the electron wave function with Ge atoms, leading to enhanced valley splitting.

Improvements for AI systems

Here are the specific improvements to AI systems that could be derived from this scientific paper, along with what those improved systems could achieve:

  1. AI-Driven Material Design and Synthesis (Generative Chemistry/Materials Informatics):

  2. AI can use the correlation between Ge concentration profiles (derived from STEM/SIMS) and quantum dot valley splitting energy to predict optimal heterostructure growth parameters (Si well thickness, interface width, barrier composition).

  3. The improved system could perform inverse design—given a target valley splitting energy in a Si/SiGe system, the AI would suggest specific atomic-resolution Ge concentration profiles (like those shown in Fig. 1(e)) required to achieve that splitting while maintaining acceptable electron mobility (low disorder).

  4. What the improved system could do: Design novel semiconductor quantum dot architectures with predictable spin qubit properties, bypassing lengthy trial-and-error experimental growth cycles.

  5. AI-Enhanced Quantum Device Simulation and Modeling (Quantum Transport Simulation):

  6. The AI can integrate the experimentally derived relationships (e.g., Fig. 4: Valley splitting energy dependence on magnetic field and disorder) into sophisticated theoretical models (like the Schrödinger-Poisson virtual crystal Hamiltonian).

  7. The improved system could perform high-fidelity simulations of spin qubit performance, accounting for both intrinsic material disorder effects and engineered interface disorder effects simultaneously, allowing for accurate prediction of initialization fidelity, control precision, and readout errors under various operational magnetic field conditions.

  8. What the improved system could do: Accurately model and predict the performance metrics (e.g., gate fidelities) of future silicon spin qubit designs before fabrication begins, significantly reducing R&D time in quantum computing hardware development.

  9. AI-Driven Performance Benchmarking and Optimization (Machine Learning for Device Characterization):

  10. The AI can analyze the comprehensive classical and quantum transport data (Mobility vs. density, Conductivity vs. density, and Magnetotransport data) across different heterostructures (A, B1–B3) to learn the complex trade-offs between increased valley splitting and decreased mobility/increased disorder.

  11. The improved system could automatically classify heterostructure performance based on a multi-objective fitness function (e.g., maximize valley splitting while keeping mobility above a threshold). It would identify which interface geometries (broad vs. sharp) offer the best low-disorder trade-off for specific qubit applications.

  12. What the improved system could do: Rapidly screen thousands of potential material growth conditions to find the optimal Si/SiGe interface geometry that balances competing physical requirements for scalable quantum architectures, accelerating the transition from theory to experimental realization.

  13. AI-Driven Disorder Analysis and Error Mitigation (Statistical Physics/ML for Noise Characterization):

  14. By analyzing the linear relationships derived in Fig. 4 (e.g., relating valley splitting energy to the overlap parameter η1), the AI can develop predictive models for how random alloy disorder specifically impacts spin qubit decoherence pathways, such as spin-valley mixing or charge noise susceptibility.

  15. The improved system could provide real-time feedback mechanisms for experimental control systems, suggesting dynamic adjustments (e.g., slight changes in gate voltages or magnetic fields) to mitigate the effects of interface disorder on qubit coherence during operation.

  16. What the improved system could do: Develop advanced error correction protocols specifically tailored to the known types of disorder present in Si/SiGe heterostructures, leading to more robust and fault-tolerant quantum computation schemes implemented on these platforms.

Sources

Related papers