We present the results from a focused ion beam instrument designed to implant single ions with a view to the fabrication of qubits for quantum technologies. The difficulty of single ion implantation is accurately counting the ion impacts. This has been achieved here through the detection of secondary electrons generated upon each ion impact. We report implantation of single bismuth ions with different charge states into Si, Ge, Cu and Au substrates, and we determine the counting detection efficiency for single ion im- plants and the factors which affect such detection efficiencies. We found that for 50 keV implants of Bi++ ions into silicon we can achieve a 89% detection efficiency, the first quantitative detection efficiency measurement for single ion implants into silicon without implanting through a thick SiO2 film. This level of counting accuracy provides implantation of single impurity ions with a success rate significantly exceeding that achievable by random (Poissonian) implantation.
Quantum computers have been proposed that exploit entangled quantum states between atoms that are isolated from environmental perturbations in a “semiconductor vacuum” which can be formed by cryogenically cooling an isotopically pure, defect free crystalline layer consisting of Si, or Ge. In a preliminary investigation of an implant and deposition layer exchange technique to produce such “vacuums”, a layer of aluminium was implanted with 28Si using a conventional implanter. After annealing and cross sectioning, layer exchange was observed to have produced multiple isolated crystals in a cross sectional TEM image. Further deposited Al layers were implanted with Ge using a SIMPLE (Single Ion Multispecies Positioning at Low Energy) implanter over a range of fluences. After anneals at 250 °C and Al removal, crystals of Ge (which also contained Si) were seen at areal densities that increased with implant fluence.
The theoretical error rates in deterministic ion implantation when using an ion beam governed by a Poisson point process with a detector that counts the impacts are investigated. It is concluded that if the error rates are small, then for spots with nominally one implanted ion the probability of failure to implant the correct number is ≈ 𝜅/𝜆+𝜂⎯⎯+𝜆/2 for a synchronous (i.e., pulsed) system or 𝐾/𝐿+𝜂⎯⎯+𝐿𝑡s for an asynchronous (i.e., continuous beam) system, where 𝜂⎯⎯ is the probability that the detector misses an ion impact, and 𝐿(𝐾) and 𝜆(𝜅) are the number of ions (dark counts) per unit time and per pulse, respectively. ts is the system reaction time for an asynchronous system. This approximation allows easy identification of the greatest need for engineering effort. Some experimental efforts to measure these parameters and their uncertainties are examined.