The spin orientation of a single nucleus is a robust, long-lived, well-isolated candidate for a scalable solid-state quantum bit. Recent experiments show that both the electron and nuclear spins of a single implanted phosphorus atom in nanoscale silicon devices can be programmed and read-out. The next step in building a quantum processor is to observe the coherent transport of a quantum state along a chain of donors. The exchange coupling interaction moves the electron (and the quantum information) from one place to another along the chain; fabricated by placing donors deterministically. This goal will need precision arrays of single dopant atoms controlled by associated nano-circuitry fabricated to a precision comparable with a few times the Bohr radius of a donor electron orbit. One way of doing this is to employ deterministic ion implantation through a scanned cantilever with a nanoscale collimator: nanostencil lithography.

A single phosphorus ion implanting into a silicon substrate with sufficient energy to penetrate 20 nm below the surface (14 keV kinetic energy) produces enough ionization to be detected with on-chip detector electrodes. An electrode pulse signals the arrival of a single implanted ion. Following this signal the nanostencil is stepped to a new site where the next ion can be implanted. In this example, the precision of the implant is limited by: the straggle of the ion implant (7–11 nm potentially less for lower energy implants and higher mass donors); the width of the the aperture (<10 nm); and the precision of the stepper drive (<1 nm). The sum of all uncertainties is less than the 30 nm required which is sufficient for the fabrication of proof-of-principle devices. The relative simplicity of the method, employing standard tools derived from scanning probe microscopy for the stepping and ion implantation potentially allow devices to be mass-produced.

In the study, nanostencils were milled with a focused ion beam and then narrowed to produce a collimating aperture by backfilling with platinum from an electron beam cracked precursor gas. The geometry of the collimator was sought by measuring the energy spectrum of high energy ions (500 keV) transmitted through the collimator. The collimator's geometric parameters were adjusted in a Monte Carlo ion beam simulation until the transmitted energy of the simulated beam fitted the experimental data. This geometry governs the rate of random ion scattering processes in a 14 keV ion implant. Scattered ions could lead to randomly placed implants in the substrate outside of the array under construction. The results show that this practical combination of techniques could readily fabricate, for example, an 8-atom linear chain with a yield of greater than 50%. Chains of donor atoms, mass produced by the technique, could be used to test ideas for coherent spin transport, counted dopant atom devices or coupled spin devices in quantum computer chips.

The researchers presented their work in the journal Nanotechnology.