Dr. David Reilly, with the help of team members from Microsoft and the University of Sydney, just developed a cryogenic quantum control platform that can control thousands of qubits at cryogenic temperatures. The platform employs CMOS circuits and is named Gooseberry. Instead of using thousands of wires in a fridge, Gooseberry operates at 100 milliKelvin (mK) and dissipates low power. The cooling power does not exceed standard refrigerators.
This research was recently published in Nature magazine earlier in January, called ‘A Cryogenic Interface for Controlling Many Qubits’. The research has also been used to create a general-purpose cryo-compute core, the first of its kind. It is one step up the quantum stack and operates around 2 K using liquid helium. This is 20 times warmer than Gooseberry and means that 400 times the cooling power is available.
Both of these chips help a large-scale quantum computer’s components communicate, as well as between the computer and a user. With these chips, not only can every qubit be used in information processing, but a stable cold environment is achieved. This is a challenge for systems containing tens of thousands of qubits, and the Microsoft team pushed past a lot of obstacles for these feats.
Even if quantum computers are measured by how many qubits they have, all of them use different types and these have unique performances. This means that comparing them is difficult. Microsoft Quantum scientists are currently developing topological qubits, and they have higher levels of error protection. This means that more power can be dedicated to useful tasks.
These qubits are the base of the quantum stack and are below the quantum-classical interface. This layer contains two layers, cryogenic control, and room temperature control. The latter is right below the classical computing level, and that contains software and quantum applications. Gooseberry sits with the qubits themselves at the same temperature as it sends instructions from the cryo-compute core as voltage signals to the qubits.
Where Gooseberry sits is important, and by being near the cold qubits, the heat might affect qubit effectiveness. The scientists placed the chip at the right position so that it does not warm the qubits up, but the heat is drawn away to a mixing chamber. This might solve the heat problem, but there is another complication. The chip must be at the same temperature as the qubits at 100 mK, and normal CMOS chips make this difficult. This is why fully-depleted silicon-on-insulator (FDSOI) technology is used in Gooseberry. This technology optimises the system so it works best at cryogenic temperatures. Gooseberry has back-gate bias, in which transistors have four terminals that compensate for temperature changes. Qubits can be calibrated individually and the transistors send the right voltage to each qubit.
Gooseberry has another advantage in its design. The electrical gates controlling the qubits are charged from a single source as necessary through the gates. Previous designs needed one-to-one cables from multiple sources at room temperature or 4K, and this prevents qubits from working at a large scale. Dr. Reilly’s team pioneered this design, and as a result, the heat is greatly dissipated with the help of cryogenic temperature. The cold lets capacitors hold charges longer. Gates no longer have to be replaced as frequently, which results in less heat and qubit instability.
Digital and analog blocks make up the Gooseberry chip. A finite-state machine (FSM) is what coupled digital logic circuits perform communication, waveform memory, and autonomous operation through. The digital part contains a master oscillator. For easier communication with the higher parts of the stack, there is a serial peripheral interface (SPI). Charge-lock fast-gate (CLFG) cells make up the analog components. The charge-lock function is for charging gates, which contain individualised voltages for each qubit. The fast-gating function is what sends the voltages to the qubits from the gate, making it directly involved in qubit information processing.
These pulses pose a challenge, namely low power dissipation. There are three variables that impact dissipation, voltage level, frequency, and capacitance. The qubit sets the voltage, and both the qubit and clock rate of the quantum plane affect the frequency. Capacitance remains as the only variable that is adjustable when charging gates and sending pulses to create low power dissipation. Tiny capacitors are spaced closely and near the quantum plane, allowing for minimal power to shuffled charges as the qubits communicate with them.
There was a test conducted by the scientists to see how Gooseberry would work with a GaAs-based quantum dot device. The device had several gates connected to a digital-analog converter (DAC) at room temperature to compare results with standard control approaches. A second quantum dot was used to measure power leakage from the CLFG cells, and the quantum dot conductance was measured to monitor the charge-locking process. The temperature during operation stayed below 100mK within the needed range of frequencies or clock speeds.
The scientists estimated the amount of power Gooseberry needs to work as a function of frequency and the number of output gates. The results take clock speeds and temperature needed by the qubits into account. After calculations, the chip was found to be able to operate within acceptable limits while it communicated with thousands of qubits. This CMOS-based approach seems to be viable for qubit platforms based on electron spins or gatemons as well.
The general-purpose cryo-compute core is a recent development that builds on Gooseberry. It can handle triggering manipulation and data. Having fewer limitations from temperature also allows it to deal with branching decision logic, and this requires more digital blocks and transistors than Gooseberry. The core works as an intermediary between Gooseberry and developer-writable executable code. This means that the qubits and outside world can communicate. It is possible to compile and execute many code types in a cryogenic environment, and this goes beyond what Gooseberry is already capable of.
Microsoft scientists believe that this is only the beginning, and more advancements are on the way to be discovered. They may take years but Microsoft is content at playing the long game.
About Microsoft Quantum
Microsoft is involved in the quantum industry, where its goal is to solve the toughest problems on the planet that quantum computing can tackle. Using world-class technology, the company is a leader in the industry. Microsoft also works with other quantum companies frequently.