Pushing the limits of technology requires pushing the limits of materials. For smaller components and circuitry, high purity is paramount. Quantum computing is a perfect example of when the details count. Fortunately, there are good engineering students paying attention to those details.
Kevin Dwyer, a University of Maryland Materials Science and Engineering graduate student is working with National Institute of Standards and Technology (NIST) researchers to enable silicon-based quantum computing. Quantum computing requires extremely high purity to avoid information loss. Qubits, which store that information, are sensitive to the spin state of the silicon atoms. An unpaired spin, causing decoherence, ruins the information.
As described in a University of Maryland press release, you can’t just buy highly pure, highly-enriched silicon. That is because it requires more than just eliminating all non-silicon atoms. The silicon itself must also be devoid of isotopes. Particularly harmful is silicon-29 (29Si), which makes up nearly 5% of unenriched silicon.
So how pure is pure enough? Four nines (99.99% of silicon-28 (28Si)) is a minimum, but Dwyer and the other researchers were not satisfied with just enough. They have achieved better than six nines (99.9999% 28Si) by their process. Just as impressive, the process is relatively simple.
To reduce the 29Si below 1 part per million they use a technique similar to mass spectroscopy. First, silane (SiH4) gas is introduced into the vacuum chamber and ionized. Silicon ions are then sent around a bend at high speed by a magnetic field. Their trajectory is dictated by their mass (i.e., their momentum). With its extra neutron, 29Si cannot turn as sharply and can be separated out.
In various iterations this purity record was reached by decreasing the vacuum pressure and increasing the distance after separation for higher selectivity. Also, in the beginning of the research they were making amorphous silicon. Although useful for testing the purification process, crystalline silicon is needed for quantum computing. To increase the crystallinity the target to which the silicon atoms were directed is heated, thereby allowing the atoms to more readily arrange into an orderly pattern.
Currently the process has been demonstrated on samples of 1 square centimeter and is specific to silicon. The team plans to extend the technique to silicon-germanium (SiGe), another important compound with similar issues. As long as they keep dedicated engineering students on the job, there is no reason to doubt they’ll be successful.
Photo courtesy of UM News/ Jenny Lee, NIST