The problem
Superconducting quantum devices use microwaves, while long-distance links work better with light. An interface must bridge them. Can it transfer a quantum signal from the processor to an optical fibre without losing too much of it or adding unwanted noise?[1]
A quantum signal leaving a superconducting processor must enter a converter, change from microwave to optical frequency and emerge at an optical port. Can the complete interface preserve useful quantum information while meeting its loss, noise, usable frequency range and heating requirements?
Conversion efficiency compares converted optical photons with microwave input photons. Internal conversion excludes losses associated with coupling through the ports. As an illustration using assumed values, take 90% internal conversion and port-coupling fractions of 80% and 50%. In the review’s two-port efficiency model,
The external conversion efficiency is 36%; these are invented inputs, not measured device results. This calculation alone says nothing about added noise or how faithfully a quantum state is preserved. An August 2026 atomic-memory experiment adds controllable retrieval timing, with efficiency normalised to the microwave input incident on the active medium. This omits spatial-overlap losses that matter at a complete hardware interface. September pump-free Bell-pair work proposes a heralded protocol, where a measurement signals success, for teleportation in a theoretical model.[1] [2] [3]
Scope
Microwave–optical interfaces for useful quantum-state transfer between superconducting hardware and optical networks, jointly controlling external efficiency, added noise, bandwidth and pump heating.
What remains open
It must transfer the quantum signal without too much loss or extra noise.[1]
Positive unassisted one-way quantum capacity in the pure-loss model requires transmissivity above one half; added noise also matters. Heralded or teleportation protocols have different requirements. Area-normalised conversion does not remove mode-matching losses at a real hardware interface.[1] [2] [3]
Why it matters
Optical links could let separated microwave-based quantum processors exchange quantum information. This requires more than a bright converted signal: loss and extra noise must leave the information usable at the receiving device.[1]
Networking depends on the delivered quantum information and the protocol that uses it. Controllable timing could coordinate distant operations, while low loss and noise constrain whether the delivered state is useful. Comparing efficiencies at the same interface boundary prevents an impressive internal conversion number from hiding port losses.[1] [2]
Definitions
3 terms
External conversion efficiency
Conversion efficiency evaluated through the input and output ports. In the cited two-port model it includes the internal conversion efficiency multiplied by the microwave and optical port-coupling fractions.[1]
Added noise
Unwanted excitations introduced during conversion, distinct from signal loss. Both must be assessed when deciding whether an interface transfers useful quantum information.[1]
Sources
- Microwave-to-optical quantum transduction of photons for quantum interconnects
Notes on Microwave-to-optical quantum transduction of photons for quantum interconnects
Review separates external port efficiency from internal conversion, and loss from added noise. The one-half capacity threshold applies to its unassisted pure-loss channel model.
- Quantum-memory-assisted on-demand microwave-optical transduction
Notes on Quantum-memory-assisted on-demand microwave-optical transduction
An atomic-memory experiment stores a microwave excitation and retrieves optical light on demand. Area-normalised conversion is distinct from a complete superconducting-network interface.
- Pump-Free Microwave-Optical Bell Pair Generation for Teleportation-Based Quantum Transduction
Notes on Pump-Free Microwave-Optical Bell Pair Generation for Teleportation-Based Quantum Transduction
Preprint proposes pump-free Bell-pair generation for teleportation-based conversion in a theoretical model; it is not an implemented network demonstration.