Quantum networks promise secure communication and distributed computing, but they lack memory that can be integrated on a chip and operate at fiber-optic wavelengths. Atomic ensembles store photons well, but they occupy an entire optical table. Solid-state alternatives either miss the telecom band or cannot release information on demand. Nearby in the archive are papers on optical traps, buffers for the quantum internet, and cooling of mechanical disks: all seek ways to hold a quantum state in a compact device. This article follows the same direction but bets on the mechanical vibrations of a nanobeam.
Inside a cryostat at 20 mK sits a silicon optomechanical crystal: an optical resonator at 1538 nm is coupled to a mechanical mode at 5314 MHz. A laser tuned to the red sideband creates a transparency window via optomechanically induced transparency (OMIT). Signal photons, generated by modulating the same laser, enter this window and are converted into phonons. To avoid heating, the write pulse is made long and weak, while the read pulse is short. Four cascaded Fabry–Perot etalons block the control light, leaving only single photons for the detector.
The memory stored a photon pulse for 150 ns and returned it on command with an internal efficiency of 4.2%. During storage, the mean phonon number in the resonator was 0.32, which is less than one quantum. When the read pulse power was reduced to a level where thermal excitation drops to 0.17, the efficiency fell to 5×10⁻⁵. The authors emphasize that the low efficiency limits only the event rate, not the purity of the quantum state, because coincidence filtering removes noise.
What sets this work apart from neighboring studies is that it demonstrates, for the first time, an absorbing memory on an optomechanical chip in the telecom band and near the quantum limit. Earlier membrane experiments remained in the classical regime, and emissive protocols were unsuitable for external photon sources. Here, the signal comes from outside, is stored in a mechanical degree of freedom, and is retrieved on demand. This bridges two fields: quantum optics and nanomechanics.
The next step is to increase storage time using acoustic shields and two-dimensional optomechanical crystals, which dissipate heat better. Integration with single-photon sources and entangled pairs is possible: coincidence filtering will remove noise and allow testing of entanglement distribution protocols. Improving the resonator's quality factor and matching bandwidths will boost efficiency by orders of magnitude.
The result will be useful in quantum repeaters and distributed quantum computing, where memory compatible with optical fibers and chips is needed.
Optimize the thermal regime and pulse parameters, add an acoustic shield to extend storage time, and test operation with a single-photon source.
For quantum memory, the trade-off between efficiency and thermal noise remains unresolved: strong control light heats the resonator, while weak light reduces the probability of writing. This work shows that phonon occupancy can be kept below unity, but at the cost of efficiency. Further progress requires materials with better thermal conductivity and bandwidth compression schemes.
This section is addressed to the author of the paper and their colleagues. Neighbouring works were found by meaning-based search across our archive, not by word overlap. We do not review: we show the field around the work.
What the work does
The authors have built an optomechanical resonator on a silicon chip that absorbs single photons in the telecom band, stores them as phonons, and then releases them on demand. The device operates at 20 mK, and thermal excitation of the mechanical mode is held below one quantum (0.32 phonons). An internal efficiency of 4.2% was demonstrated for a storage time of 150 ns.
Where its strength is
This is the first demonstration of an absorbing quantum memory on an optomechanical chip in the telecom band with thermal excitation below one quantum. It paves the way for integrating quantum memory into photonic chips.
Where it could go next
Worth trying
Why this may matter
Such memory could become a key component of scalable quantum networks, compatible with existing fiber-optic infrastructure and silicon chip technology.
Nearby works in our archive
The archive is sparse around this work: the closest review is only 0.58, against a usual 0.64. It looks like few have drilled nearby.
The mechanical oscillation frequency of the resonator, 5314 MHz, is a hundred thousand times higher than that of a tuning fork. At 20 mK, such an oscillator contains on average no thermal phonons at all.
In the TV series 'The Expanse', quantum entanglement is used for instantaneous communication across the Solar System. Such networks require memory that can hold a photon until confirmation arrives that the other half of the pair has been delivered.