Quantum computers that are built from tiny chips made up of silicon-germanium use qubits, which, like binary for traditional computers, are the basic units of quantum information. In these chips, qubits are often based on the spin of a single hole – a ‘missing electron’ in the material – trapped inside a nanoscale electrostatic trap called a quantum dot.
For quantum computers to be of any use, the qubits need to talk to each other – whether that’s to their immediate neighbours, or to the other qubits that are all the way across a chip.
Currently, the easiest way to connect two qubits on a silicon-germanium substrate is to put them close enough to each other that the particles can tunnel between one qubit and the other. Useful if you only want short-distance communication. But if you really want the bits to interact over a longer distance – which is required for error correction and building large-scale machines – then you need a long-distance link.
There have been different ways people have attempted to enable this long-distance communication, such as physically shuttling the charge across the chip, using sound waves on the surface of the material, or using microwave signals in the metal cavities. However, these methods take up a lot of physical space, and they don’t scale well if millions of qubits need to be packed together.
But what if the charge didn’t need to be moved around or big microwave hardware wasn’t necessary? What if it went off vibes? Sound vibes.
A sound idea
Researchers from the University of Warwick and the National Research Council of Canada have proposed a new quantum chip architecture. The chip uses sound-like vibrations to enable qubits to communicate over long distances. These sound vibrations would travel inside the chip itself to link the qubits together. This is something the team call Quantum Phononic Link (QPL).
Normally, if you want to move information a long way using sound, you need to build a separate mechanical structure to carry it, plus a piezoelectric converter to translate between electricity and sound.
The material germanium, when compressed into silicon, makes hole-based qubits extremely sensitive to being physically stretched or squeezed. So, if a sound wave passes through the material, it stretches and squeezes the crystal lattice as it moves through. The squeeze directly nudges the qubit’s spin state without the need for a converter to turn the sound into an electrical signal first. The coupling is direct.
The reason the team can “trap” the sound is because of the materials they use. Germanium carries sound more slowly than the silicon-germanium surrounding it. The speed difference between the outer silicon-germanium and the germanium means that the sound wave is confined to a very thin layer – the same layer in which the qubit lives. Essentially, by changing the material’s acoustic properties, the team were able to trap the sound where they needed it.
A strong coupling
Both the sound and the qubits are confined to the exact same sliver of material. The effect of the sound waves squeezing and the qubits’ location are almost perfectly aligned, giving them a strong coupling.
Running computer simulations, the team demonstrated that by shaping this thin layer of germanium into narrow channels, they could guide the sound waves in a controlled, one-directional path between two distant qubits. They found that they could also design small cavities – dead-end pockets to trap and store the vibration, rather than just allowing it to pass through.
What is its potential?
Sound’s wavelength at any given frequency is much shorter than that of light or electrical signals because it travels much more slowly than light or electrical signals -hundreds of nanometres instead of centimetres. This means that the entire sound-guiding structure can be made tiny, which enables these links to be densely packed together on a chip with very little wasted space.
The manufacturing process is the same as the process used for ordinary computer chips, meaning that in theory, this approach should be able to scale up to full-size silicon wafers that connect qubits which are extremely close together or ones that are spread across the entire chip.
However, the team has noted that material imperfections such as rough surfaces, disordered atoms, or crystal defects will eventually scatter and weaken the sound wave the farther it travels. The researchers believe that, based on today’s materials, the sound-based links will work reliably over distances from a few dozen micrometres up to a few centimetres, and further improvements in fabrication could push that distance even farther.
The team estimates that the coupling strength would be somewhere between 0.1 and 10 million cycles per second (0.1-10MHz), which is comparable to, and in some cases better than what’s achieved with existing surface-sound-wave or microwave methods, whilst taking up less space.
The team believe that this method could become a practical way to connect distant qubits on large-scale quantum chips, and could even serve as a bridge for linking different kinds of quantum hardware together in the future. So far, this is all based on simulation and theory – the next step will be to actually fabricate and test one of these chips.
