Substrate surface engineering for high-quality silicon/aluminum superconducting resonators

Abstract

Quantum bits (qubits) with long coherence times are an important element for the implementation of medium- and large-scale quantum computers. In the case of superconducting planar qubits, understanding and improving qubits' quality can be achieved by studying superconducting planar resonators. In this Paper, we fabricate and characterize coplanar waveguide resonators made from aluminum thin films deposited on silicon substrates. We perform three different substrate treatments prior to aluminum deposition: One chemical treatment based on a hydrofluoric acid clean, one physical treatment consisting of a thermal annealing at 880 degree Celsius in high vacuum, one combined treatment comprising both the chemical and the physical treatments. We first characterize the fabricated samples through cross-sectional tunneling electron microscopy acquiring electron energy loss spectroscopy maps of the samples' cross sections. These measurements show that both the chemical and the physical treatments almost entirely remove native silicon oxide from the substrate surface and that their combination results in the cleanest interface. We then study the quality of the resonators by means of microwave measurements in the "quantum regime", i.e., at a temperature T~10 mK and at a mean microwave photon number $\langle n_{\textrm{ph}} \rangle \sim 1$. In this regime, we find that both surface treatments independently improve the resonator's intrinsic quality factor and that the highest quality factor is obtained for the combined treatment, $Q_{\textrm{i}} \sim 0.8$ million. Finally, we find that the TLS quality factor averaged over a time period of 3 h is $\sim 3$ million at $\langle n_{\textrm{ph}} \rangle \sim 10$, indicating that substrate surface engineering can potentially reduce the TLS loss below other losses such as quasiparticle and vortex loss.

Similar papers