A beam of visible light enters a cloud of billions of rubidium atoms cooled to near absolute zero, and emerges on the other side already in the telecom range. That is the work of a translator that does not lose the meaning of the phrase: the atoms receive a photon in the "language" of 795 nanometers and emit it in the "language" of 1367 nanometers, suitable for traveling through optical fiber. The translation itself takes less than a microsecond, but it is what decides whether the quantum internet will work.

Quantum networks connect remote nodes via optical fiber, but the best candidates for quantum memory, atoms, emit in the visible range, where losses in glass are huge. To transmit a quantum state over distance, it must be converted to the telecom range without destroying it. In this work, the translator is a cold cloud of rubidium-87 trapped in a magneto-optical trap. Three laser beams, a probe at 795 nm, a coupling beam at 780 nm, and a control beam at 1324 nm, are focused into the cloud, where a diamond-scheme four-wave mixing process (a nonlinear process where three waves generate a fourth) produces a signal photon at 1367 nm.

Previously, atomic systems for frequency conversion lagged behind solid-state crystals in efficiency, and the record 65% in a cold ensemble was not backed by detailed experimental data. Now it is shown that the diamond scheme in cold rubidium yields 66% and 80% at optical depths of 75 and 110, respectively, surpassing all previous atomic results. The key trick is choosing Zeeman sublevels (energy states of the atom split by a magnetic field) that form a cyclic transition, eliminating population losses and preserving coherence. Parameter optimization was guided by measured electromagnetically induced transparency spectra (a quantum effect that makes the medium transparent to the probe beam) of two embedded configurations, V-type and cascade.

This translator can already serve as an interface between atomic quantum memory and an optical fiber link in prototypes of quantum repeaters. The next step is to replace the weak coherent probe field with a true single-photon source and directly test the preservation of quantum purity of the converted state. The theoretical model predicts that at an optical depth of 200, the efficiency will approach 90%. Then atomic nodes will be able to exchange photons in the telecom range without losing quantum information, bringing the dream of a quantum internet one step closer.