Scalable Atomic Arrays for Spin-Based Quantum Computers in Silicon.

Semiconductor spin qubits combine excellent quantum performance with the prospect of manufacturing quantum devices using industry-standard metal-oxide-semiconductor (MOS) processes. This applies also to ion-implanted donor spins, which further afford exceptional coherence times and large Hilbert space dimension in their nuclear spin. Here multiple strategies are demonstrated and integrated to manufacture scale-up donor-based quantum computers. 31PF2 molecule implants are used to triple the placement certainty compared to 31P ions, while attaining 99.99% confidence in detecting the implant. Similar confidence is retained by implanting heavier atoms such as 123Sb and 209Bi, which represent high-dimensional qudits for quantum information processing, while Sb2 molecules enable deterministic formation of closely-spaced qudits. The deterministic formation of regular arrays of donor atoms with 300 nm spacing is demonstrated, using step-and-repeat implantation through a nano aperture. These methods cover the full gamut of technological requirements for the construction of donor-based quantum computers in silicon.

Deterministic Shallow Dopant Implantation in Silicon with Detection Confidence Upper-Bound to 99.85% by Ion-Solid Interactions.

Silicon chips containing arrays of single dopant atoms can be the material of choice for classical and quantum devices that exploit single donor spins. For example, group-V donors implanted in isotopically purified 28 Si crystals are attractive for large-scale quantum computers. Useful attributes include long nuclear and electron spin lifetimes of 31 P, hyperfine clock transitions in 209 Bi or electrically controllable 123 Sb nuclear spins. Promising architectures require the ability to fabricate arrays of individual near-surface dopant atoms with high yield. Here, an on-chip detector electrode system with 70 eV root-mean-square noise (≈20 electrons) is employed to demonstrate near-room-temperature implantation of single 14 keV 31 P+ ions. The physics model for the ion-solid interaction shows an unprecedented upper-bound single-ion-detection confidence of 99.85 ± 0.02% for near-surface implants. As a result, the practical controlled silicon doping yield is limited by materials engineering factors including surface gate oxides in which detected ions may stop. For a device with 6 nm gate oxide and 14 keV 31 P+ implants, a yield limit of 98.1% is demonstrated. Thinner gate oxides allow this limit to converge to the upper-bound. Deterministic single-ion implantation can therefore be a viable materials engineering strategy for scalable dopant architectures in silicon devices.