Towards Robust Automation of Quantum Dot Bootstrapping

The tuning and management of quantum dot qubits is a large and complex task. The tools used for automated tuning schemes vary from simple fittings to heuristically defined algorithms to traditional computer vision techniques. A host of machine-learning-based techniques have also been utilized. However, most of the tuning efforts focused on the more advanced phases of tuning, assuming that the device is already pre-tuned, with a properly calibrated charge sensor and that the "safety" regimes for all gates are already known. However, the initial phase of tuning – the so-called "device bootstrapping" – is still nearly always done heuristically, requiring a highly trained researcher to be responsible for the subsequent decisions on how to adjust the relevant parameters. We develop an automated routine to bridge the gap between the initial device cool-down and a voltage configuration in which other previously developed automation schemes can take over for a multiple quantum dot device. The result of this autotuning procedure provides a sufficient starting point for the wide-ranging set of tasks for control of quantum dot qubits. The initial device bootstrapping routine can also be used as a device-screening process to determine which devices should be used for qubit formation and which devices should be discarded.