Researchers at Delft University of Technology have designed germanium heterostructures that achieve up to three orders of magnitude increase in spin-orbit interaction (SOI), a critical property for advancing spin-based quantum devices. The team proposes enriching unstrained germanium channels with localized, strained silicon spikes, a departure from current designs that rely on compressively strained germanium. They used multi-objective Bayesian optimization to maximize SOI while maintaining compatibility with existing manufacturing processes. These novel heterostructures also predict GHz-scale spin splittings for hybrid superconducting Andreev spin qubits, potentially opening pathways to scalable quantum and spintronic applications.
Unstrained Ge Channels Enhance Spin-Orbit Interaction
This substantial enhancement addresses a critical limitation in developing high-performance spin qubits and hybrid superconducting quantum devices, where strong SOI is essential for efficient spin control. The heterostructure design uses localized strain to manipulate electron behavior, offering a pathway beyond the constraints of traditional strain-relaxed silicon-germanium barriers. The optimization process specifically accounted for potential variations during growth, ensuring the resulting heterostructures are robust and reproducible.
This approach yielded designs where the SOI coefficient, β2, reaches values exceeding 50 nanometers, a figure significantly higher than previously attainable in germanium-based systems. Researchers found that the optimal spike placement, with distances of 4.8 and 11.3 nanometers, delivered maximal SOI under applied electric fields of 1.5 millivolts per nanometer.
The predicted increase in SOI translates directly to improved quantum-dot spin qubit quality factors, with the new heterostructures expected to yield performance up to two orders of magnitude better than current materials. “In hybrid superconducting-semiconducting architectures, large intrinsic SOI ensures strong coupling to superconducting resonators, efficient manipulation of superconducting (Andreev) spin qubits, and topological phases of matter,” the paper states, highlighting the broad implications of this advancement.
By incorporating localized silicon spikes into unstrained germanium, they effectively engineer the electronic band structure to enhance spin-orbit coupling. Simulations demonstrate that these heterostructures maintain a substantial HH-LH gap of approximately 3.28 meV, enabling spin splitting up to 1.2 meV before the anti-crossing point.
Robustness against growth imperfections was a key consideration throughout the design process. Analysis reveals that the optimized heterostructures maintain a high SOI coefficient, greater than 40 nanometers, even with variations of ±0.5 nanometers in the positioning of the silicon spikes. This resilience is critical for practical device fabrication, minimizing the impact of unavoidable imperfections in the epitaxial growth process. The researchers also included a bottom silicon-germanium barrier, positioned 30 nanometers below the top barrier, to further enhance the HH-LH gap and improve device performance.
The optimized designs, detailed in accompanying documentation, demonstrate a significant improvement over unstrained germanium, achieving a 40-fold increase in SOI. This represents a leap forward in materials science for quantum technologies, offering a pathway to overcome limitations previously imposed by weak spin-orbit interaction in germanium-based devices. The work demonstrates that carefully tailored heterostructures, designed with the aid of machine learning, can allow for new ways of manipulating and controlling spin in quantum systems.
Localized Silicon Spikes Optimized via Bayesian Optimization
The process focused on optimizing the shape of localized silicon additions within unstrained germanium channels. This approach yielded heterostructures with significantly enhanced SOI, a property vital for efficient spin control in quantum systems. The optimization process simultaneously addressed potential instabilities caused by electric-field fluctuations, maximizing both SOI strength and device robustness. Parameters like spike distance, electric field, and interface width were all refined through the Bayesian optimization, resulting in profiles that balance performance and manufacturability.
The team modeled the silicon concentration profile at each interface using a logistic function, allowing precise control over the spike’s shape and position. This enhancement stems from the increased SOI, which allows for faster and more precise manipulation of electron spins.
The optimization process also considered the impact of variations in well width, confirming the robustness of the designs even with slight deviations from ideal growth conditions. The team also investigated the relationship between SOI and the energy gap, finding that larger SOI regions correlate with smaller energy gaps, a key consideration for device performance.
“We identify multiple routes to engineer Si gradients – implemented via a single Si-poor bump and two Si-rich spikes – and leverage multi-objective Bayesian optimization to chart the Pareto front of heterostructures displaying optimal trade-offs between SOI maximization, and its robustness against parameter variations,” the researchers state. The robustness of these heterostructures is further demonstrated by their resilience to variations in well width, maintaining performance even when the width exceeds 25 nanometers. The optimization process successfully identified numerous optimal solutions, as visualized in supplementary documentation, demonstrating the efficiency of the machine learning approach.
The team’s work builds on recent advances in growing unstrained germanium channels with strained silicon-germanium barriers, providing a new method for enhancing SOI and unlocking the potential of germanium-based quantum technologies. The combination of advanced materials design and machine learning optimization has yielded a significant leap forward in the field.
Ge+ Heterostructure Design Predicts GHz-Scale Andreev Qubit Splittings
A multi-objective Bayesian optimization process was central to this advancement, specifically designed to maximize SOI while simultaneously ensuring compatibility with existing epitaxial growth techniques and maintaining robustness against realistic variations in growth parameters. This optimization didn’t simply target high SOI, but also considered the practicalities of manufacturing, a critical step toward viable quantum technologies. Analysis of a hybrid superconducting-semiconducting system reveals that the newly designed germanium heterostructures, dubbed “Ge+”, significantly outperform unstrained germanium in Andreev spin qubit applications.
A recent analysis already showed unstrained germanium providing larger spin splittings than epsilon-germanium, but Ge+ promises even greater gains. The team’s calculations indicate that Ge+ consistently achieves spin splittings exceeding 100 MHz, and even enters the GHz regime, a threshold important for efficient microwave control and readout of superconducting spin qubits.
This is achieved through the manipulation of the energy levels within the heterostructure, specifically by exploiting the anti-crossing behavior with the light-hole subband. The relationship between chemical potential and spin splitting was carefully modeled, revealing a non-monotonic behavior in Ge+ that differs from both germanium and epsilon-germanium. Specifically, the team found that the spin splitting reaches a maximum near a chemical potential of 1 meV before decreasing at higher values, a detailed effect that requires precise control of material composition.
“Ge+ show consistently spin splittings above 100 MHz, entering also the GHz regime,” the researchers report, highlighting the potential for significantly faster qubit operation. This behavior is linked to the influence of the light-hole subband, which creates a unique energy landscape within the heterostructure. The work builds on parallel research focused on strain engineering of germanium Andreev spin qubits, with collaborative discussions and insights exchanged between research groups.
The researchers acknowledge contributions from V. Coppini, A. Manesco, A. Akhmerov, V. Fatemi and B. Van Heck, emphasizing the collaborative nature of the research. Funding for the project came from The Netherlands Ministry of Defense and the U.S. Army Research Office, with a disclaimer noting that the views expressed are those of the authors and do not necessarily reflect the policies of these organizations. The team’s detailed manual optimizations on distances maximizing , as visualized in supplementary figures, further validate the predicted performance enhancements.
k·p Model Details Rashba Spin-Orbit Interaction in Ge
A detailed k·p model reveals the mechanisms driving enhanced spin-orbit interaction (SOI) in newly designed germanium heterostructures, predicting a significant increase in the Rashba SOI strength. The model accounts for the complex interplay of hole subbands and material composition, demonstrating how localized silicon spikes within unstrained germanium channels can overcome limitations of conventional designs. Specifically, the analysis focuses on the impact of engineered silicon concentration gradients on the cubic in momentum Rashba SOI, parameterized by lengths β₂ and β₃, which dictate the strength of spin manipulation.
The theoretical framework employs a six-band k·p Hamiltonian, incorporating effects from finite momentum, epitaxial strain and the virtual crystal approximation to model the silicon concentration gradients. This approach allows for a precise calculation of the spin-splitting energy, Eso(k), at various crystal momenta, providing insights into the energy landscape experienced by spin qubits.
Calculations reveal that the SOI is substantially modified by the spatial variation of material constants, particularly at interfaces and within the engineered silicon spikes. The model identifies two primary contributions to β₂; one arising from broken spatial inversion symmetry and a second, novel term originating from the spatial variation of material constants within the heterostructure. The researchers found that state-of-the-art Ge heterostructures show negligible intrinsic SOI due to large epitaxial strain fields.
This equation highlights how the spatial distribution of hole wave functions and the gradient of the Luttinger parameter γ₂ contribute to the overall SOI strength. By carefully tailoring the silicon concentration profile, the team was able to maximize this effect, achieving up to three orders of magnitude enhancement in SOI compared to standard compressively strained germanium structures.
The model predicts that the optimized designs yield a spin splitting energy that scales with the cube of the wavevector, indicating a strong dependence on momentum and enabling efficient electrical control of spin. The optimization process considered nine structural parameters, including the thickness and amplitude of a silicon bump, the barrier-bump distance, silicon content in the barriers, residual strain, applied electric field and interface broadening.
The model demonstrates that a smooth silicon-poor bump, combined with two sharp silicon spikes, provides the most effective pathway to enhance SOI. By solving the full six-band k·p Hamiltonian via the envelope function approximation, the team accounted for higher-order corrections to the spin-splitting energy, ensuring the accuracy of their predictions at higher momenta.
The detailed analysis of the energy dispersion, E(ky), for the optimized heterostructure confirms the enhanced SOI and its potential for realizing high-performance spin qubits. The work demonstrates that engineering silicon concentration spikes in germanium provides an efficient tool for coherent spin control.
Quantum Dot Quality Factors Improve with Engineered Si Profiles
This enhancement, detailed in recent work, addresses a longstanding limitation in germanium-based quantum devices, weak spin-orbit interaction, by moving beyond designs reliant on compressively strained germanium. This is critical because strong spin-orbit interaction ensures efficient coupling to superconducting resonators, a necessary component for controlling and manipulating quantum information. The resulting heterostructures demonstrate a substantial improvement in device performance, with predicted quality factors exceeding those of current materials by up to two orders of magnitude.
The researchers state that the design relies on creating spatial variation in material constants within the heterostructure, specifically through the engineered silicon concentration gradients. This variation induces a Rashba spin-orbit interaction, a phenomenon where the spin of an electron precesses in momentum space, enabling electrical control of spin degrees of freedom. Detailed analysis of the energy dispersion confirms this prediction, revealing a clear enhancement of spin-orbit interaction in the optimized heterostructures.
The researchers employed a six-band k·p Hamiltonian within the envelope function approximation to accurately model the electronic structure of the heterostructures, accounting for heavy-hole and light-hole subband interactions. These predictions are significant because they suggest the potential for faster qubit manipulation and improved coherence times, essential for building practical quantum computers. The team also investigated the impact of realistic growth imperfections, demonstrating the robustness of their designs against variations in material composition and interface width. The researchers anticipate that these heterostructures will enable the development of more powerful and versatile quantum devices.
👉 More information
🗞 Tailoring germanium heterostructures for quantum devices with machine learning
✍️ Patrick Del Vecchio, Kevin Rossi, Giordano Scappucci and Stefano Bosco
🧠 DOI: http://link.aps.org/doi/10.1103/r1v9-mff5




See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.
