Cryo-CMOS Antenna for Wireless Communications within a Quantum Computer Cryostat

Scaling quantum computers from a few qubits to large numbers remains one of the critical challenges in realizing practical quantum advantage. Multi-core quantum architectures have emerged as a promising solution, enabling scalability through distributed quantum processing units (QPUs) interconnected via classical and quantum links. However, the bottleneck of wired connections persists, as densely packed wired interconnects, both vertically across temperature stages and horizontally within the same layer, introduce spatial constraints, power dissipation, and latency, which could hinder performance as the number of QPUs increases. To overcome these limitations, this work proposes a cryo-compatible on-chip differential dipole antenna operating at 28 GHz to enable short-range wireless communication within a quantum computer cryostat. Temperature-dependent material properties are incorporated to accurately capture antenna behavior at 4 K. Moreover, by embedding the antenna in a realistic cryostat structure, we evaluate the feasibility of antenna operation within the cryogenic environment. The proposed antenna achieves a reflection coefficient of -20.8 dB in free space and -18.38 dB within the cryostat, demonstrating efficient impedance matching.

💡 Research Summary

The paper addresses a critical bottleneck in scaling quantum computers: the massive wiring required to control and read out thousands of qubits. Dense interconnects across multiple temperature stages introduce spatial constraints, power dissipation, thermal load, and latency, all of which threaten the delicate cryogenic environment needed for high‑fidelity qubit operation. To mitigate these issues, the authors propose an on‑chip differential dipole antenna that can operate at 4 K and enable short‑range wireless communication between quantum processing units (QPUs) inside a cryostat.

The antenna is a half‑wave dipole realized in a standard cryo‑CMOS process. Two copper traces form the dipole arms on a SiO₂ dielectric, fed differentially by a pair of microstrip lines terminating in Ground‑Signal‑Signal‑Ground (GSSG) pads. This differential feeding suppresses common‑mode currents, improving noise immunity and matching the requirements of low‑power cryogenic RF transceivers. The design starts from the classic resonance condition L = λ₀/(2√ε_eff) and is refined through full‑wave electromagnetic simulations (CST MWS) that incorporate temperature‑dependent material parameters. At 4 K, copper conductivity rises from 5.9 × 10⁷ S/m (room temperature) to 2.9 × 10⁸ S/m, while silicon’s conductivity drops dramatically, both effects reducing dielectric and ohmic losses.

Key geometric parameters after optimization are: dipole arm length L_opt ≈ 2.8 mm, arm gap S_opt ≈ 0.03 mm, top metal (M7) thickness 3.5 µm, SiO₂ thickness 3.823 µm, and silicon substrate thickness 0.30 mm. Simulations in free space show a reflection coefficient S₁₁ of –28.85 dB at 28 GHz (cryogenic) and –21.81 dB at room temperature, corresponding to radiation efficiencies of 92 % and 83 % respectively.

To evaluate realistic performance, the antenna is embedded in a model of a typical dilution‑refrigerator cryostat (30 cm diameter, 70 cm height) that includes multiple temperature stages, thermal shielding, and an outer Mu‑metal magnetic shield. The antenna is placed on a PCB above a cooling plate, and the dipole length is slightly increased to 3.06 mm to compensate for the altered surrounding dielectric. Inside the cryostat the simulated S₁₁ is about –21 dB, the total efficiency drops to 77 % (due to interaction with the metallic walls and dielectric layers), and the realized gain reaches 8.25 dBi at 28 GHz. Field plots confirm the 180° phase shift between the two arms and reveal multi‑path propagation within the enclosure, which could cause interference if not properly managed.

The paper’s contributions are threefold: (1) a detailed methodology for designing a cryogenic on‑chip differential dipole antenna that accounts for temperature‑dependent material properties; (2) a comprehensive electromagnetic analysis of the antenna both in free space and within a full cryostat model, demonstrating viable impedance matching and respectable gain; (3) quantitative evidence that such an antenna can achieve –20 dB or better reflection, >70 % radiation efficiency, and >8 dBi gain at 28 GHz, making it a promising candidate for wireless inter‑QPUs links.

Future work suggested includes mitigating common‑mode conversion caused by thermal gradients, exploring antenna arrays for beamforming and spatial multiplexing, and integrating the antenna with a cryo‑CMOS transmitter/receiver chain to validate actual data‑rate, bit‑error‑rate, and latency performance. If successfully realized, wireless links based on this antenna could dramatically reduce wiring density, lower thermal load, and improve scalability of multi‑core quantum computers.