Optomechanical resource for fault-tolerant quantum computing

Fusion-based quantum computing with dual-rail qubits is a leading candidate for scalable quantum computing using linear optics. This paradigm requires single photons which are entangled into small resource states before being fed into a fusion network. The most common sources for single optical photons and for small entangled states are probabilistic and heralded. The realization of a single reliable deterministic source requires many redundant probabilistic sources and a complex optical network for rerouting and retiming probabilistic outputs. In this work, we show how optomechanics enables reliable production of resources for photonic quantum computing without the redundancy of the all-optical approach. This is achieved by using acoustic modes as caches of quantum resources, ranging from single-particle states to small entangled states, with on-demand read-out. The advantages of acoustic modes as optical quantum memories, compared to other technologies, include their intrinsically long lifetimes and that they are solid state, highly tailorable, and insensitive to electromagnetic noise. We show how the resource states can be prepared directly in the acoustic modes using optical controls. This is still probabilistic and heralded, as in the all-optical approach, but the acoustic modes act as a quantum memory which is integrated into the production of the states. The quantum states may be deterministically transferred from acoustic modes to optical modes, on demand, with another optical drive.

💡 Research Summary

The paper addresses a central bottleneck in linear‑optics quantum computing (LOQC): the deterministic supply of single photons and small entangled resource states. Conventional approaches rely on either deterministic emitters (quantum dots, atoms) that still suffer from indistinguishability and efficiency issues, or probabilistic photon‑pair generation followed by heralding. The latter requires many redundant sources and fast, low‑loss optical switches to multiplex the heralded photons into a deterministic stream, which dramatically increases system complexity and introduces loss and timing challenges.

The authors propose a fundamentally different architecture based on cavity optomechanics. In their scheme, acoustic (phonon) modes of a mechanical resonator serve as long‑lived quantum memories (“caches”) that can store the quantum resource after a probabilistic, heralded preparation step, and release it on demand as an optical photon. The core device consists of four optical modes—designated blue, herald, red, and output—coupled to a single mechanical mode. Two pairs of optical modes are used for distinct interactions: a two‑mode squeezing interaction (blue–herald) creates photon‑phonon entanglement, while a beam‑splitting interaction (red–output) swaps the phonon excitation into the output optical mode.

During the preparation phase, a strong blue‑detuned pump drives the blue–herald pair, implementing a parametric‑down‑conversion‑like Hamiltonian that creates correlated photon‑phonon pairs. Detection of a photon in the herald mode heralds the successful creation of a single phonon in the mechanical resonator. Because the mechanical quality factor can be extremely high (Γ≈10 Hz), the phonon can be stored for seconds, far exceeding the timescales of optical routing. When a photon is needed, a red‑detuned pump activates the beam‑splitting interaction, coherently converting the stored phonon into a photon in the output mode, which can be coupled into a waveguide for downstream processing.

The authors present two parameter regimes. The “target” set reflects the best‑in‑class experimental numbers: optical loss κ≈1 GHz, mechanical frequency Ω≈10 GHz, vacuum coupling g₀≈100 kHz, and acoustic linewidth Γ≈10 Hz. The “near‑term” set uses more readily achievable bulk acousto‑optic resonators: κ≈50 MHz, g₀≈1 kHz, Γ≈50 Hz. Numerical simulations show that, under target parameters, single‑photon generation fidelity exceeds 99 % with indistinguishability above 99 %, while even the near‑term parameters achieve >90 % fidelity and >95 % indistinguishability. Importantly, the architecture scales: many identical optomechanical resonators can be operated in parallel, each preparing a phonon independently. Because the phonon storage is essentially loss‑free on the relevant timescale, the probability that all resonators have a ready phonon before any read‑out is near unity, enabling massive parallelism without the exponential overhead typical of pure heralded schemes.

Beyond single photons, the paper outlines how small entangled states (Bell pairs, three‑qubit GHZ, small cluster states) can be generated. Multiple resonators are prepared simultaneously; subsequent beam‑splitting operations and post‑selection on herald detectors create the desired entanglement across different optical modes. The mechanical cache again guarantees that the entangled phonons survive until the optical conversion stage, eliminating the need for fast optical switching during state preparation.

Implementation challenges are discussed in detail. Precise frequency spacing (Ω) between the optical mode pairs must be maintained to avoid unwanted two‑mode squeezing or beam‑splitting with spurious modes. The authors propose mode hybridization, use of orthogonal transverse profiles (e.g., TE₀₀ vs TE₀₁), or opposite propagation directions to suppress undesired couplings. They also compare macro‑scale resonators (which tolerate higher pump powers and have lower optical absorption) with nano‑scale devices (which offer larger g₀ but suffer from heating and surface loss). The analysis favors macroscopic resonators for the present application because they can be strongly driven without degrading the mechanical Q‑factor, enabling the effective coupling rates needed for high‑efficiency conversion.

In summary, the work demonstrates that an optomechanical system can act as a deterministic, on‑demand source of both single photons and small entangled photonic states, while sidestepping the massive multiplexing overhead of purely optical approaches. By leveraging the long coherence time of acoustic modes as quantum memories, the scheme provides a scalable pathway toward fault‑tolerant LOQC, potentially simplifying hardware architecture, reducing latency, and improving overall resource efficiency. The authors argue that this approach could become a cornerstone for future large‑scale photonic quantum processors.