Growth and Characterization of Epitaxial Al Layers on GaAs and Si Substrates for Superconducting CPW Resonators in Scalable Quantum Computing Systems

摘要

The growth of Aluminum (Al) on semiconductors and dielectrics is a cornerstone in the quest for scalable quantum computing systems. Indeed the electrical properties of Al make it an exceptional candidate for the realization of superconducting resonators, pivotal tools for understanding and operating superconducting qubits. Such resonators have been fabricated recently on Sapphire substrates, using molecular beam epitaxy (MBE), and displayed quality factors above a million. Complementary studies of these resonators have demonstrated that the metal-substrate interface was the primary source of decoherence and losses, highlighting the importance of pristine interfaces (free of contaminants), and high quality epitaxial growth in order to minimize the native defects level. In this work we investigate different substrate materials in order to yield equivalent or higher quality factor resonators. Gallium Arsenide (GaAs) and Silicon (Si) were selected for their good dielectric properties, well-established processing techniques and a potential on-chip integration. After thermal substrate annealing, and in some cases deposition of a buffer structure, Al was grown on both substrates at low temperature, using MBE. In view of the extreme sensitivity of the resulting Al crystal orientation to the initial surface conditions, different starting surface reconstructions were investigated. Growth evolution was studied with reflection high energy electron diffraction simultaneously at several azimuths during deposition on rotating substrates. The substrate temperature, the system background pressure and possible sources of contamination were monitored carefully to ensure the reproducibility of the results. Resulting layers were subsequently characterized with X-ray diffraction (XRD) to confirm their epitaxial nature and crystallographic orientation. Finally, atomic force microscopy was used to assess the layers morphology. Different growth modes were observed depending on the material: Al grew in a Stranski-Krastanov mode on GaAs(001) surfaces, in a Frank-van-der-Merwe mode on Si(111) surfaces and in a Volmer-Weber mode on Si(001) surfaces. All yielded crystalline structures. Targeting atomically smooth single crystalline materials, best results were obtained for Al(110) deposited on GaAs(001)-(2x4) substrates with surfaces showing RMS roughness of 0.552nm. While the epitaxy on Si(111)-(``1x1") led to single-crystalline Al(111) layers with a RMS roughness of only 0.487nm, a detailed XRD study indicated a possible misalignment of the crystallites that could induce defects in the material. Similarly, epitaxy on Si(111)-(7x7) substrates yielded Al(111) layers of a RMS roughness of 0.519nm that, however, appeared rougher under the Nomarski microscope, likely due to the surface preparation prior to Al deposition. The deposition on Si(001)-(2x1) substrates led to bi-crystals of Al(110) of higher RMS roughness (0.719nm). Finally, the growth on GaAs(001)-(4x4) reconstruction led to polycrystalline materials with mixed Al(100), Al(110) and Al(111) of RMS roughness 1.20nm. Moreover, the composition of the layers grown on the GaAs(001)-(4x4) reconstruction was inconsistent across multiple growths.