A scalable and high performance approach to readout of silicon qubits
Abstract
Fast, high fidelity qubit readout that can be scaled to large numbers of qubits is an important ingredient for error corrected quantum computers. It is also of high value for demonstration experiments on the way to that ultimate goal. The currently most successful approach for semiconductor spin qubits uses RF readout of charge sensors, which involves bulky RF components that are cumbersome to install already at the ten-qubit level, and do not offer a clear scaling pathway to much larger qubit counts. First results indicate that baseband readout using transistors placed near the qubit chip as first amplifier state is a promising alternative. We propose to improve the performance of charge sensors adjacent to the qubit by exploring an innovative sensing dot design that promises a much larger output signal than conventionally possible. The key idea of our sensing dot concept is to spatially separate the drain lead by using a large bias, which leads to a much reduced dot-drain capacitance. As a result, large voltage swings on the drain can be achieved without affecting the dot potential, thus avoiding the feedback effect that limits the signal of conventional sensing dots. Gate-defined few-electron double quantum dot devices with such modified sensors will be realized in Si/SiGe heterostructures. The performance of this concept is intimately related to that of the transistor. Hence, we also aim to improve the letter by testing low-power transistors for use at low temperature, and by reducing their input capacitance to enhance the bandwidth. Our estimates indicate that near-shot-noise-limited single shot readout with very high fidelities at microsecond-scale averaging times should be possible while maintaining a power dissipation that would allow the operation of many thousands of parallel readout channels in a conventional dilution refrigerator. Using our technique, we will demonstrate and characterize single shot readout for the most prominent spin to charge conversion schemes, namely by Pauli spin blockade and energy-resolved tunneling to the reservoir, thus covering single electron qubits as well as multi-electron qubit encodings. We will examine the achievable fidelity and readout time in relation to the amplifier power consumption. To understand possible difficulties with this approach arising from the high bias voltage, the back action of charge sensors will be studied as a function of operating condition. Both relaxation of the qubit during the readout phase and dephasing of the qubit during manipulation, when the charge sensor is not needed, will be considered, the primary experimental technique for the latter will be spin echo based dephasing measurements.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 11, 2018
- Source ID
- W911NF1710349
Entities
People
- Hendrik Bluhm
Organizations
- Army Contracting Command
- National Security Agency
- RWTH Aachen University