学术文献库

Silicon spin qubits stand out due to their very long coherence times, compatibility with industrial fabrication, and prospect to integrate classical control electronics. To achieve a truly scalable architecture, a coherent mid-range link that moves the electrons between qubit registers has been suggested to solve the signal fan-out problem. Here, we present a blueprint of such a $\approx 10\,μ$m long link, called a spin qubit shuttle, which is based on connecting an array of gates into a small number of sets. To control these sets, only a few voltage control lines are needed and the number of these sets and thus the number of required control signals is independent of the length of this link. We discuss two different operation modes for the spin qubit shuttle: A qubit conveyor, i.e. a potential minimum that smoothly moves laterally, and a bucket brigade, in which the electron is transported through a series of tunnel-coupled quantum dots by adiabatic passage. We find the former approach more promising considering a realistic Si/SiGe device including potential disorder from the charged defects at the Si/SiO$_2$ layer, as well as typical charge noise. Focusing on the qubit transfer fidelity in the conveyor shuttling mode, we discuss in detail motional narrowing, the interplay between orbital and valley excitation and relaxation in presence of $g$-factors that depend on orbital and valley state of the electron, and effects from spin-hotspots. We find that a transfer fidelity of 99.9 \% is feasible in Si/SiGe at a speed of $\sim$10 m/s, if the average valley splitting and its inhomogeneity stay within realistic bounds. Operation at low global magnetic field $\approx 20$ mT and material engineering towards high valley splitting is favourable for reaching high fidelities of transfer.