Geometric scaling of dielectric loss in superconducting coplanar waveguides

Typ
Examensarbete för masterexamen
Master's Thesis
Program
Nanotechnology (MPNAT), MSc
Publicerad
2023
Författare
Emil, Rehnman
Modellbyggare
Tidskriftstitel
ISSN
Volymtitel
Utgivare
Sammanfattning
For superconducting qubits, coplanar waveguide (CPW) resonators play an imprtant role in qubit readout and control. They can also be used as a probe for loss in qubits. Loss in CPWs is mainly caused by two-level systems (TLS), radiation and quasi-particles. These competing loss sources are all affected by the resonator size. By measuring the power dependence of the internal quality factor (Qi) of resonators with varying geometry we find that for narrow resonators, TLS loss is limiting low-power Q, and radiation is the main source of high-power loss. Wide resonators show a weaker Qi power dependence, which suggests a smaller contribution from TLS loss. Changing the thin film deposition method greatly improved Qi for these resonators indicating that quasi-particle loss was limiting Qi. Another parameter that affects TLS loss is the gap between the CPW centre conductor and ground plane. Using a large gap, we measured single-photon level Qi of 1.5 million. The larger gap reduces the TLS loss by diluting the electric filed in the TLS hosting regions. We also investigate the loss in so-called trenched resonators. By removing the silicon substrate in the resonator gap, one of the TLS hosting regions can be moved away from the areas with the highest electric field strengths. A trenching recipe using deep reactive ion etching was developed, and trenched resonators were fabricated and measured. The low-power quality factor in the trenched resonators was significantly higher compared to the standrad resonator, and we achieved a single-photon level Qi of 1.9 million.
Beskrivning
Ämne/nyckelord
superconducting qubits, quantum computing, dielectric loss
Citation
Arkitekt (konstruktör)
Geografisk plats
Byggnad (typ)
Byggår
Modelltyp
Skala
Teknik / material
Index