Raman Scattering in Atomic Ensembles
- Raman Scattering in Atomic Ensembles is an inelastic process where incident photons interact with atoms to generate frequency-shifted light and coherent spin-wave excitations.
- It underpins quantum interfaces by enabling efficient quantum memory, multimode storage, and entanglement generation through controlled light–matter interactions.
- Spatial mode engineering and careful management of decoherence are key to optimizing performance in applications like parametric amplification and random lasing.
Raman scattering in atomic ensembles refers to inelastic light scattering processes where incident photons interact with atoms, producing frequency-shifted photons (Stokes or anti-Stokes) while coherently transferring population and phase information between long-lived atomic states. This phenomenon plays a critical role in light–matter quantum interfaces, quantum memories, nonlinear optics, and coherent spectroscopy. Its study encompasses the dynamics of collective atomic excitations ("spin waves"), spatial and temporal multimode structure, decoherence mechanisms, and parametric correlations between photons and matter. Both spontaneous (vacuum-seeded) and stimulated (coherence-seeded) Raman regimes underpin various applications, including efficient quantum entanglement distribution, state conversion, and high-gain nonlinear amplification.
1. Theoretical Formalism of Raman Scattering in Atomic Ensembles
The prototypical system consists of a Λ-type three-level atom with two long-lived ground states (, ) and one excited state (). Under large single-photon detuning Δ (off-resonant regime), the excited state can be adiabatically eliminated. The effective interaction Hamiltonian couples the strong classical "write" (pump) field, the quantized Stokes field, and the ground–ground (spin-wave) coherence:
Here, is the Raman coupling constant for each spatial mode , while and create a Stokes photon and a collective spin-wave excitation respectively. In real space, this becomes:
with the nonlinear susceptibility, being dipole matrix elements, and the classical and quantum field components (Chrapkiewicz et al., 2014).
The collective spin-wave mode with wavevector is:
where the sum runs over all atoms in the ensemble. The spontaneous process admits a two-mode squeezing solution, and under undepleted-pump approximation, the joint state of Stokes photons and spin waves is entangled (Kolodynski et al., 2012, Bian et al., 2011).
2. Multimode Structure and Spatial Mode Engineering
The spatial degree of freedom leads to a natural multimode decomposition of the Raman interface. The finite transverse size (waist ) and cell length () define a discrete set of orthonormal spatial modes, often Laguerre–Gaussian (LG) or Hermite–Gaussian, for both the scattered light and the spin wave. For each pair of modes, the effective Hamiltonian reduces to independent two-mode squeezers:
The number of significantly occupied spatial modes is set by the Fresnel number (with the pump wavenumber); experimentally up to $15-16$ in warm vapor cells (Chrapkiewicz et al., 2014).
Mode-specific decay rates, determined by atomic diffusion and dephasing, set the temporal and spatial bandwidths, and dictate the multimode storage and retrieval capacity of atomic quantum memories (Kolodynski et al., 2012, Parniak et al., 2015).
3. Storage, Retrieval, and Decoherence Dynamics
After preparation, the spin-wave excitation decays predominantly via diffusion-induced dephasing (, where is the diffusion constant), and possible homogeneous broadening :
When a read (retrieval) pulse is applied, the stored spin-wave coherence is mapped back onto an anti-Stokes field (or Stokes field, depending on the phase-matching and detuning). The expected retrieved photon number in mode after storage time is:
where is the intrinsic retrieval efficiency. The spatial and temporal mode structure remain correlated due to phase-matching imposed by the geometry and the Doppler effect (Chrapkiewicz et al., 2014, Dąbrowski et al., 2014).
4. Parametric and Correlation Properties
Raman scattering in atomic ensembles implements a multimode parametric amplifier: spontaneous emission creates Stokes–spin-wave pairs with strong correlations. The process can be interpreted as delayed four-wave mixing, with the sequence "write → storage → read" forming a closed four-photon process:
Second-order intensity correlations for Stokes (S) and anti-Stokes (A) modes quantify the degree of quantum correlation:
indicating nonclassical pair correlations well above thermal statistics (Chrapkiewicz et al., 2014, Bian et al., 2011). The phase of the Stokes field and spin wave are strongly anti-correlated—direct phase-sum measurements confirm this, underpinning continuous-variable EPR entanglement between light and matter.
5. Enhanced, Stimulated, and Correlation-Seeded Regimes
Beyond spontaneous vacuum seeding, Raman gain can be dramatically enhanced in two ways:
- Coherently enhanced Raman scattering: Preparing a macroscopic atomic spin wave (via a weak "write-1" pulse) before the main Raman interaction leads to collective, phase-matched emission, described by the enhancement factor:
where is the density of spin-flipped atoms, the gain kernel (Yuan et al., 2010). Precise spatial and phase matching between the read pulse and the stored coherence is essential to maximize gain enhancement.
- Correlation-enhanced scattering: If a light field initially correlated with the atomic spin wave is injected, interference leads to phase-sensitive control of the Raman gain. Constructive relative phase yields gain far exceeding the sum of uncorrelated (light-seeded and spin-wave-seeded) contributions. This effect can be used to implement SU(1,1) interferometry and quantum-enhanced metrology (Yuan et al., 2012).
6. Experimental Implementations and Applications
Raman scattering in warm atomic ensembles (notably 87Rb vapor) forms the basis for high-bandwidth, multimode quantum memories, photon-pair sources, and nonlinear parametric amplifiers. Experimental parameters typical of high-efficiency implementations include:
- Cell: 100 mm × 25 mm, , optical depth ~130, buffer gas (Kr or Ne) to control diffusion.
- Write/read detunings: ±1 GHz from respective ground–excited state transitions.
- Storage/retrieval times: microseconds, determined by the buffer gas and diffusion constants.
- Number of spatial modes: up to ≈16, limited by the Fresnel number and diffusion-induced decoherence.
- Measured correlations: single-shot images clearly show conjugate-direction speckle correlations between Stokes and anti-Stokes photons (Chrapkiewicz et al., 2014, Dąbrowski et al., 2014).
Applications extend to:
- Quantum repeater protocols and network nodes, leveraging light–matter entanglement and efficient retrieval.
- Generation of continuous-variable multipartite entanglement by exploiting collective spin waves as "entanglers" under electromagnetically induced transparency (EIT) (Yang et al., 2013).
- Phase-sensitive, nonlinear interferometry and high-resolution spectroscopy using correlation-seeded Raman amplification (Yuan et al., 2012).
- Frequency conversion and broadband squeezed light generation, tuning the anti-Stokes–Stokes balance by Hamiltonian detuning control (Parniak et al., 2015, Dąbrowski et al., 2014).
7. Advanced Topics: Disorder, Radiation Trapping, and Lasing Thresholds
In disordered atomic ensembles (e.g., dense alkali vapor), multiple scattering and radiation trapping on closed transitions (e.g., F₀=3 → F=4 in Rb) permit diffusive transport of Raman-scattered photons and the possibility of random lasing. The macroscopic susceptibility and ladder-approximation Bethe–Salpeter equations govern amplification vs. losses, including inelastic anti-Stokes channels. The threshold for random lasing is set by the product of the gain length and transport mean free path:
where is the gain length, the scattering mean free path, and the system size (Gerasimov et al., 2014). Control of polarization and Zeeman degeneracy enables fine-tuning of the gain/loss balance in such systems.
For further technical details, analytic derivations, and illustrative experimental data, see (Chrapkiewicz et al., 2014, Yuan et al., 2010, Yuan et al., 2012, Parniak et al., 2015, Dąbrowski et al., 2014, Kolodynski et al., 2012, Bian et al., 2011, Yang et al., 2013), and (Gerasimov et al., 2014).