Optimisation of electrically-driven multi-donor quantum dot qubits

Abstract

(Dated: Multi-donor quantum dots have been at the forefront of recent progress in Si-based quantum computation. Among them, 2 P : 1 P qubits have a built-in dipole moment, making them ideal for electron dipole spin resonance (EDSR) using the donor hyperfine interaction, and thus all-electrical spin operation. The development of all-electrical qubits requires a full understanding of their EDSR and coherence properties, in which multi-donor dot qubits are expensive to model computationally due to the multi-valley nature of their ground state. Here, by introducing a variational effective mass wave-function, we examine the impact of qubit geometry and nearby charge defects on the electrical operation and coherence of 2 P : 1 P qubits. We report four outcomes: (i) The difference in the hyperfine interaction between the 2 P and 1 P sites enables fast EDSR, with T π ∼ 10 − 50 ns and a Rabi ratio ( T 1 /T π ) ∼ 10 6 . We analyse qubits with the 2 P : 1 P axis aligned along the [100], [110] and [111] crystal axes, finding that the fastest EDSR time T π occurs when the 2 P : 1 P axis is k [111], while the best Rabi ratio occurs when it is k [100]. This difference is attributed to the difference in the wave function overlap between 2 P and 1 P for different geometries. In contrast, the choice of 2 P axis has no visible impact on qubit operation. (ii) Sensitivity to random telegraph noise due to nearby charge defects depends strongly on the location of the nearby defects with respect to the qubit. For certain orientations of defects random telegraph noise has an appreciable effect both on detuning and 2 P − 1 P tunneling, with the latter inducing gate errors. (iii) The qubit is robust against 1 /f noise provided it is operated away from the charge anticrossing. (iv) Entanglement via exchange is several orders of magnitude faster than dipole-dipole coupling. These findings pave the way towards fast, low-power, coherent and scalable donor dot-based quantum computing.