Automated extraction of quantum dot energy levels from pulsed-gate spectroscopy data

The extraction of energy levels in the dots is an important component of tuning quantum dots as qubits for quantum computing. A common experimental method to determine such energy levels is pulsed-gate spectroscopy, performed with charge sensing as a function of pulse amplitude, such as in Dodson et al. (2022). Lines in such data correspond to energy levels. Here we report an automated analysis of such data. We combine a simple peak-finding function applied across the entire data set with physics-informed heuristics to find line-like features characteristic of pulsed-gate spectroscopy in data. In addition to eliminating the need for human input, these methods enable analysis of data with significantly less optimization of the signal-to-noise ratio than is typically required, decreasing the requirements on experimental measurement. We demonstrate real-time analysis of this data in a Si/SiGe double quantum dot device. We also compare this approach with several machine learning algorithms, focusing on the speed and accuracy of each method. 

Dodson, J.P., Ercan, H.E., Losert, M.P. et al. How valley-orbit states in silicon quantum dots probe quantum well interfaces. Phys. Rev. Lett. 128, 146802 (2022).