Cryogenic mixed-signal readout electronics for quantum computers
More Info
expand_more
Abstract
Quantum computers promise large speedups compared to classical computers for specific classes of problems by exploiting quantum phenomena for computation. Despite notable progress towards larger systems, today's quantum computers still lack the necessary size for realizing many of the anticipated benefits. Scaling to these larger numbers is complicated by the fragility of the qubits, which raises the necessity of operating the quantum computing core at deep cryogenic temperatures to limit the influence of noise. However, in the present experimental setups for quantum computing the electronic interfaces are often based on equipment operated at room temperature. This arrangement creates a potential wiring bottleneck towards cryogenic temperature, that is connected to reliability limitations. A proposed solution to this challenge is moving the interfacing electronics close to the qubits into the cryogenic environment. However, implementation of the electronics at cryogenic temperature places a set of system constraints that prevents directly porting the room-temperature electronics. The most prominent of these constraints is the limited cooling power at deep cryogenic temperature, requiring the interfacing electronics to save power wherever possible. To implement the necessary functions of the interface in practice, CMOS technology is uniquely positioned due to its large scale integration capabilities, low-power digital logic and sustained performance at cryogenic temperature.
In this work, contributions to the readout electronics of quantum computers are made, with an emphasis on the readout of semiconductor spin quantum computers. Readout poses a significant challenge for current spin-qubit systems, especially its scalable integration into the electronic interface. This dissertation can be separated into contributions to the readout of quantum computers in three main categories: cryogenic digitization, cryogenic noise characterization and readout benchmarking.
The wide-band signals present in the frequency-multiplexed RF readout interface require corresponding wide-band digitization capabilities of the associated data converters. In this work, we present robust, power-efficient and high-speed time-interleaved SAR ADC designs adopted to operation at cryogenic temperature to serve this application. These include the first high-speed cryo-CMOS SAR ADC suitable for operation in the RF readout interface, further efficiency improvements to this ADC and finally an ADC tightly integrated with an efficient body-bias enabled dynamic pre-amplifier. The body-bias enabled pre-amplifier also demonstrated excellent room-temperature performance, with the best reported combined power efficiency of a high-speed SAR and driver circuit.
All designs have been realized in a 40nm CMOS technology and characterized at cryogenic temperature.
For the directed design of many of the functions in the electronic interface, accurate models for simulation are required. While significant work has been presented on DC characterization at cryogenic temperatures, comparatively little work covering low-frequency noise is available. To help further development, an extensive low-frequency noise characterization is performed on a 40nm CMOS technology at cryogenic temperature. The results of this characterization allow a broad overview over the low frequency noise behavior of the technology. Given the radical change in temperature between room-temperature and cryogenic temperature, a surprisingly small overall change in the input-referred noise is observed. The most prominent difference is the finding of a systematic Lorentzian feature at cryogenic temperature that is for the first time described here.
Lastly, while promising designs have been suggested for the DC-readout of spin qubits, guidelines for directing such designs have been lacking. In order to provide a set of such guidelines, we first demonstrate that the limits of DC-readout lie significantly beyond current state-of-the-art, with much room for improvement on the electronics. Then, we analyze the voltage amplifier, the transimpedance amplifier, the charge sampling, and the current pre-amplifier by deriving their design equations and trade-offs.
Finally the various architectures are compared with the fundamental limit and suggestions for specific use-cases are give. The results suggest that DC-readout is a promising path for the development of a scalable spin-qubit readout.
Files
File under embargo until 25-05-2025
File under embargo until 25-05-2025