IN Brief:
- Krypton sputtering produced body-centred cubic tantalum on silicon at temperatures as low as 200°C.
- Conventional direct deposition has largely required temperatures above 400°C to obtain the same tantalum phase.
- The researchers fabricated transmon qubits with quality factors reaching 16.9 million while retaining a back-end-compatible process window.
Cornell University researchers have demonstrated a sputtering process that forms the preferred body-centred cubic phase of tantalum on silicon at substrate temperatures as low as 200°C. The method replaces argon with krypton as the sputtering gas, widening the process window for superconducting tantalum films that have otherwise often required direct-deposition temperatures above 400°C.
The work addresses a fabrication problem rather than a shortage of promising superconducting materials. Tantalum has produced high-performance microwave resonators and qubits, but its useful crystal phase can be difficult to deposit directly on silicon within the temperature limits of established semiconductor process flows. At lower temperatures, tantalum can form a different phase with less desirable electrical properties, while higher processing temperatures increase the risk of intermixing at the tantalum-silicon interface.
Postdoctoral researcher Maciej Olszewski proposed changing the sputtering gas to alter the energy transferred during deposition. Krypton atoms are heavier than argon atoms, so the sputtering process can eject tantalum atoms with greater momentum. Cornell’s team found that the change promotes formation of body-centred cubic tantalum on silicon at 200°C without requiring the higher substrate temperature normally associated with direct growth of that phase.
The temperature reduction is significant because many semiconductor fabrication lines impose thermal budgets on later process steps. Once temperature-sensitive metals, dielectrics, or previously patterned structures are present, another excursion above 400°C may no longer be practical. A process that achieves the required material phase at 200°C fits more comfortably within back-end-of-line constraints and leaves more margin for process variation.
Material quality did not collapse with the lower temperature. The Nature Materials paper reports substantially higher electronic conductivity in the krypton-sputtered films and a tight distribution of microwave performance in coplanar-waveguide resonators. Films grown at higher temperatures showed greater microwave loss correlated with increased tantalum-silicon intermixing, reinforcing the point that simply increasing substrate temperature is not an unlimited route to better material.
The researchers also fabricated transmon qubits from the films. The published work includes devices with a compact 20 µm capacitor gap and reports quality factors reaching 16.9 million. Those results place the material process firmly in device territory rather than stopping at structural or resistivity measurements, although the figures come from an academic fabrication environment rather than a high-volume foundry flow.
Improving the tantalum layer also exposed sensitivity elsewhere in the process. Cornell’s team found that device performance had become increasingly dependent on subtle changes in formation of the Josephson junction, the nonlinear element used to make the transmon qubit. As losses in the surrounding materials are reduced, process steps that were previously masked can become the next limiting factor.
That progression is familiar in semiconductor manufacturing. Removing one dominant source of variation rarely finishes the process; it shifts attention to interfaces, contamination, lithography, deposition uniformity, junction formation, or packaging that previously contributed less to the final result. Quantum hardware adds exceptionally low operating temperatures and coherence requirements to the usual need for repeatability.
The study was published in Nature Materials on 18 August, following earlier preprint work from the team. Its relevance to manufacturing lies in the combination of performance and a broader process window. A material that works only under a narrow laboratory recipe is difficult to transfer into production, however impressive the individual device result.
Krypton sputtering does not solve the wider scaling problem for superconducting quantum processors. Junction reproducibility, control wiring, interconnect, packaging, cryogenics, and error correction remain substantial engineering constraints. It does remove some of the tension between high-performance tantalum and semiconductor-compatible thermal budgets, giving process engineers more room to integrate the material without forcing the rest of the wafer flow to accommodate a temperature step above 400°C.
