Engineers in the UK and Switzerland have developed a method for reading several qubits at the same time to scale up quantum computers.
A team of engineers in EPFL in Switzerland worked with researchers from the Hitachi Cambridge Laboratory as well as Quantum Motion Technologies and the University of Cambridge on a millikelvin integrated circuit fabricated using 40nm CMOS technology that integrates silicon quantum-dot arrays. The key is the support electronics in an architecture that allows the array to be efficiently addressed and read.
The architecture contains integrated microwave lumped-element resonators for dispersive sensing of the charge state of the quantum dots, mediated via digital transistors in a column–row-addressing distribution. The chip demonstrates combined time- and frequency-division multiplexing, which scales sublinearly the resources as well as footprint required for readout.
- $8m for UK CMOS cryo chip
- Microsoft cryogenic CMOS chip controls thousands of qubits
- Intel taps Dutch team for second generation cryo controller
“IBM and Google currently have the world’s most powerful quantum computers,” said Prof. Edoardo Charbon, head of the Advanced Quantum Architecture Laboratory (AQUA Lab) in EPFL’s School of Engineering which led the work. “IBM has just unveiled a 127-qubit machine, while Google’s is 53 qubits.”
“Our challenge now is to interconnect more qubits into quantum processors – we’re talking hundreds, even thousands – in order to boost the computers’ processing power,” he said. The number of qubits is currently limited by the fact that there’s no technology yet available that can read all the qubits rapidly. “Complicating things further, qubits operate at temperatures close to absolute zero, (50mK),” said Charbon. “That makes reading and controlling them even harder. What engineers typically do is use machines at room temperature and control each qubit individually.”
“Our method is based on using time and frequency domains,” said Andrea Ruffino, a PhD student at Charbon’s lab. “The basic idea is to reduce the number of connections by having three