Performance of Academic and Industrial Spin-1/2 Qubits in ²⁸Si/SiGe

Spin-qubit research, historically spearheaded by academic institutions, is undergoing a significant transformation as the semiconductor industry takes a more active role, opening exciting opportunities for scaling up spin-based quantum hardware. However, despite this growing involvement, direct comparisons between academic and industrial spin-qubit devices remain scarce in the literature. This study compares silicon spin-qubit devices with micromagnets, fabricated in RIKEN’s academic cleanroom facilities and Intel’s 300mm industrial foundry, evaluating key performance metrics such as coherence times, gate fidelities, and noise characteristics. The underlying academic (TU Delft) and industrial (Intel) ²⁸Si/SiGe heterostructures were both grown by reduced-pressure chemical vapor deposition, using ²⁸SiH₄ as a gas precursor to achieve isotopically enriched Si-28 quantum wells with 800 ppm residual isotopes and similar strain levels.

Both device types achieve single-qubit gate fidelities well above 99.9%. However, academic devices, benefiting from faster operation speeds, achieve even higher fidelities, reaching almost 99.999% in isolated operation and over 99.99% in simultaneous operation of up to three qubits. Industrial devices, however, show longer T₂^Hahn times, suggesting reduced high-frequency noise, likely due to advanced manufacturing techniques.

Noise in academic devices primarily arises from charge noise caused by two-level systems (TLS) in the device oxides, leading to correlated noise between neighboring qubits due to shared TLS. In contrast, the noise experienced by the qubits in the industrial device is lower and dominated by residual nuclear spins in the host material, resulting in minimal correlated noise. This provides a critical advantage for mitigating correlated quantum errors, one of the most challenging aspects of quantum error correction (QEC).

Industrial spin qubits offer better scalability and, with further refinements, could surpass academic devices in performance. These findings highlight the current state of the field and underscore the potential of industrial fabrication techniques to enhance the scalability of spin-based quantum hardware.