Citation: Kane, Bruce E. "A silicon-based nuclear spin quantum computer." nature 393.6681 (1998): 133-137.
Web: https://www.nature.com/articles/30156
Tags: Physical, Hardware, Foundational
This review is part of a homework assignment for a quantum information science class.
This paper introduces a method for quantum computation based on the nuclear spins of an array of phosphorus atoms in a silicon chip. This sort of computer is now known as the Kane quantum computer. Silicon is chosen as the bulk material because it is a type IV semiconductor, and thus has nuclear spin I=0 isotopes. This means that the silicon chip's nuclear spins will not interact with the nuclear spins on the phosphorus atoms. Phosphorus is chosen because it is the open nuclear spin 1/2 shallow donor in silicon. More specifically, the isotope of phosphorus one must work with is 31P.
The spins are manipulated by exploiting the hyperfine interaction and electron-mediated nuclear spin interaction. Namely, one can apply voltages to metallic gates in the device to induce magnetic fields, resulting in controlled gates on the nuclear spins.
The gates come in two kinds. The first kind is placed above the phosphorus donors, controlling the hyperfine interaction and hence the resonance frequency of the spins beneath them. The second kind is placed between the donors to turn on and off the electron-mediated coupling between the nuclear spins. The first kind allows for one qubit gates, and the second kind allows for two qubit gates.
By the end of the paper, methods are given to perform each step of the quantum computation process. This seems like a very solid proposal for a quantum computer. It needs to operate at low temperatures and high magnetic fields (T=100mK, B=2T), but the coherence times are high.
It has turned out that the most difficult part of this proposal is controlling individual magnetic fields precisely, without those magnetic fields leaking over to the surrounding phosphorous atoms. This doesn't mean that people aren't trying to make a computer this way. It has become the leading approach of quantum computation in Australia. However, more work needs to be done and it is not at all clear that this approach can be made viable.
In recent years, however, there has been a bit of renewed interest in this approach. If instead of using I=1/2 donors one uses I=7/2 donors, localized electric fields in the device can be used to coherently manipulate the nuclear spins:
> Asaad, Serwan, et al. "Coherent electrical control of a single high-spin nucleus in silicon." Nature 579.7798 (2020): 205-209.
Seeing as electric fields are easier to localize than magnetic fields, this may give now hope to the field. Of course, if I=7/2 then the resulting quantum system will have eight levels instead of two. This is not a fundamental problem, because qubits aren't important for quantum computing. The important feature is having complete control over quantum information in some system of any dimension. Working with eight levels will be more difficult for several reasons (i.e. G=Z8 error correction is much harder than G=Z2 error correction) but it could certainly be done.