Abstract
Motivated by applications to quantum computer architectures, we study the change in the exchange interaction between neighbouring phosphorus donor electrons in silicon due to the application of voltage biases to surface control electrodes. These voltage biases create electrostatic fields within the crystal substrate, perturbing the states of the donor electrons and thus altering the strength of the exchange interaction between them. We find that control gates of this kind can be used to either enhance or reduce the strength of the interaction by an amount that depends both on the magnitude and orientation of the donor separation.
Introduction
Phosphorus atoms embedded in silicon are leading candidates for qubits in solid-state quantum computing due to their exceptionally long coherence times. A fundamental requirement for a scalable quantum processor is the ability to perform multi-qubit operations, which relies on the exchange coupling ($J$) between adjacent donor electrons. However, $J$ in silicon is highly sensitive to the exact atomic placement of donors due to the complex valley structure of the silicon conduction band, leading to rapid spatial oscillations. This research investigates the use of surface electrodes, or "J-gates," to provide a controllable "knob" for tuning this interaction.
Device Architecture and Methodology
The study utilizes a theoretical model based on the Kane architecture. In this setup, phosphorus donors are typically buried at a depth of approximately 20nm within a silicon substrate, positioned beneath a thin layer of silicon oxide (5nm) and a metallic J-gate.
To calculate the exchange coupling, the researchers employed a multi-step theoretical approach:
- Wave-function Modeling: Donor electron states were described using the Kohn-Luttinger form, incorporating both the periodic Bloch functions of the silicon crystal and a hydrogenic envelope.
- Electrostatic Simulation: The Poisson equation was solved to determine the precise electric field distribution generated by the J-gate bias.
- Exchange Calculation: The Heitler-London approximation was used to find the energy difference between the spin-singlet and spin-triplet states, utilizing a basis expansion of 140 states to ensure accuracy under varying bias conditions.
Tuning the Exchange Interaction
The primary mechanism for control is the modification of the potential barrier between donors. A positive voltage bias on the J-gate lowers this barrier, pulling the electron wave-functions toward each other and increasing their spatial overlap, which significantly strengthens the exchange coupling. Conversely, a negative bias increases the barrier, effectively suppressing the interaction.
As shown in the figures above, the exchange coupling can be tuned across several orders of magnitude. For a donor separation of approximately 11nm, the coupling varies from roughly $10^{-3}$ meV to nearly 1 meV as the gate bias ranges from -1V to +1V. This wide dynamic range is crucial for implementing high-fidelity quantum gates and for isolating qubits when they are not interacting.
Orientation and Placement Sensitivity
A key finding of the study is the extreme sensitivity of the exchange coupling to the crystallographic orientation of the donor pair. The interaction is most robust and strongest when the donors are aligned along the [100] axis. In contrast, alignments along the [110] or [111] directions result in weaker coupling and increased sensitivity to small placement errors.
Even with the control provided by the J-gate, atomic-scale precision remains a challenge. The research indicates that shifting a donor by just a single lattice site can alter the coupling strength by over 50%. Furthermore, while the gate bias allows for tuning, it also tends to amplify the inherent spatial oscillations of $J$, making precise fabrication even more critical.
Conclusion and Future Outlook
This work demonstrates that surface J-gates are an effective means of controlling the exchange interaction in phosphorus-doped silicon quantum computers. While providing orders-of-magnitude tunability, the model also highlights the stringent requirements for donor placement and the breakdown of simple approximations in certain regimes, such as large negative biases. Future research will likely focus on more complex gate geometries and the integration of these control mechanisms into larger-scale qubit arrays.
Summary updated