We report an experiment in which a light pulse is decelerated and trapped in a vapor of Rb atoms, stored for a controlled period of time, and then released on demand. We accomplish this storage of light by dynamically reducing the group velocity of the light pulse to zero, so that the coherent excitation of the light is reversibly mapped into a collective Zeeman (spin) coherence of the Rb vapor.

We experimentally demonstrate emission of two quantum-mechanically correlated light pulses with a time delay that is coherently controlled via temporal storage of photonic states in an ensemble of rubidium atoms. The experiment is based on Raman scattering, which produces correlated pairs of spin-flipped atoms and photons, followed by coherent conversion of the atomic states into a different photon beam after a controllable delay. This resonant nonlinear optical process is a promising technique for potential applications in quantum communication.

Searches for extrasolar planets using the periodic Doppler shift of stellar spectral lines have recently achieved a precision of 60 cm s −1 (ref 1), which is sufficient to find a 5-Earth-mass planet in a Mercury-like orbit around a Sun-like star. To find a 1-Earth-mass planet in an Earthlike orbit, a precision of ∼ 5 cm s −1 is necessary. The combination of a laser frequency comb with a Fabry-Pérot filtering cavity has been suggested as a promising approach to achieve such Doppler shift resolution via improved spectrograph wavelength calibration 2−4 , with recent encouraging results 5 . Here we report the fabrication of such a filtered laser comb with up to 40-GHz (∼ 1-Å) line spacing, generated from a 1-GHz repetition-rate source, without compromising long-term stability, reproducibility or spectral resolution. This wide-line-spacing comb, or 'astro-comb', is well matched to the resolving power of high-resolution astrophysical spectrographs. The astro-comb should allow a precision as high as 1 cm s −1 in astronomical radial velocity measurements.The accuracy and long-term stability of state-of-theart astrophysical spectrographs are currently limited by the wavelength-calibration source 6,7 , typically either thorium-argon lamps or iodine absorption cells 8 . In addition, existing calibration sources are limited in the red-tonear-IR spectral bands most useful for exoplanet searches around M stars 9 and dark matter studies in globular clusters 10 . Iodine cells have very few spectral lines in the red and near-IR spectral bands, while thorium-argon lamps have limited lines and unstable bright features that saturate spectrograph detectors. Recently, laser frequency combs 11 have been suggested as potentially superior wavelength calibrators 2,3 because of their good longterm stability and reproducibility, and because they have useful lines in the red-to-near-IR range. The absolute optical frequencies of the comb lines are determined by f = f ceo + m × f rep , where f rep is the repetition rate, f ceo is the carrier-envelope offset frequency and m is an integer. Both f rep and f ceo can be synchronized with radio-frequency oscillators referenced to atomic clocks. For example, using the generally available Global Positioning System (GPS), the frequencies of comb lines have long-term fractional stability and accuracy of better than 10 −12 . For the calibration of an astrophysical spectrograph, fractional stability and accuracy of 3 × 10 −11 are sufficient to measure a velocity variation of 1 cm s −1 in astronomical objects. In addition, using GPS as the absolute reference allows the comparison of measurements at different observatories.For existing laser combs, f rep is usually < 1 GHz (ref. 12), which would require a spectrograph with a resolving power of R = λ/δλ 10 5 to resolve individual comb lines (here δλ is the smallest difference in wavelengths that can be resolved at wavelength λ). In practice, astrophysical spectrographs tend to have a resolving power of R ∼ 10 4 − 10 5 owing to physical limitations on the in...