Intel Spin Qubits: Automated low-temperature measurement and statistical data analysis for improved fabrication and device design

Spin qubits in silicon are in many ways similar to state-of-the-art transistor devices, and they can both be manufactured with today’s 300mm equipment and processes. The 300mm fab infrastructure is a powerful tool because of its excellent process control that results in high-quality devices with high reliability; however, it requires feedback from device performance grounded in statistical analysis of large data sets. Similarly, for quantum dots, automation and statistical data analysis of low-T measurements can be used to address challenges such as minimizing charge noise and TLS’s in dielectrics and other on-chip materials, as well as eliminating spurious dots. Here we present an automated approach to tuning up quantum dots and extracting gate crosstalk, charging energy (E_c), leverarm, charge noise, and other device performance metrics, which are then fed back to device design and the fabrication process for iterative improvements in performance. Automation, and the large data sets that result, not only enable high-throughput measurements for rapid fabrication improvements, but also allow characterization of device limitations in new ways, including identifying regions of badness and the location of TLS’s, and ultimately will point the way toward high-performance qubits.