An automated approach for consecutive tuning of quantum dot arrays

Recent progress has shown that the dramatically increased number of parameters has become a major issue in tuning of multi-quantum dot devices. The complicated interactions between quantum dots and gate electrodes cause the manual tuning process to no longer be efficient. Fortunately, machine learning techniques can automate and speed up the tuning of simple quantum dot systems. In this letter, we extend the techniques to tune multi-dot devices. We propose an automated approach that combines machine learning, virtual gates and a local-to-global method to realize the consecutive tuning of quantum dot arrays by dividing them into subsystems. After optimizing voltage configurations and establishing virtual gates to control each subsystem independently, a quantum dot array can be efficiently tuned to the few-electron regime with appropriate interdot tunnel coupling strength. Our experimental results show that this approach can consecutively tune quantum dot arrays into an appropriate voltage range without human intervention and possesses broad application prospects in large-scale quantum dot devices.