Collisional decoherence during writing and reading quantum states

Collisions, even though they do not limit the lifetime of quantum information stored in ground-state hyperfine coherences, may severely limit the fidelity for quantum memory when they happen during the write and read process. This imposes restrictions on the implementation of Raman-type quantum processes in thermal vapor cells and their performance as a quantum memory. We study the effect of these collisions in our experiment.

Figure 1

(Color online) (a) Experimental setup: two control lasers (write and read) are combined at a Wollastone prism (WP) and intersect inside a magnetically shielded ${}^{87}\mathrm{Rb}$ vapor cell with Neon buffer gas. The weak signal fields (Stokes and anti-Stokes) are separated from the control beams at a second polarizer. After spectral filtering, the signal photons are detected by fiber coupled avanlanche photo diode (APD). (b) Simplified level scheme: a $\lambda$ system is used to map ground-state coherences to the signal fields via two detuned control beams. (c) Laser pulse timing for spontaneous creation of ground-state coherence and delayed readout.

Figure 2

(Color online) Left: frequency resolved scan of the Stokes channel with imperfectly filtered write light as a frequency reference. The length of a Fabry-Perot interferometer (FP1, $\mathrm{FSR}=5\mathrm{GHz}$) was scanned while the signal intensity (bottom) and its cross correlation to the anti-Stokes channel (top) were recorded. The model (solid red line) includes the coherent $\left(S\right)$ and incoherent $\left(F\right)$ contributions to the Stokes signal and the write laser light $\left(L\right)$ that appears again two FSR away from its original frequency. The data (black dots) show that only the coherent Stokes contribution is correlated to the anti-Stokes signal. Statistical uncertainties in the measured cross correlation were calculated according to the finite number of experimental cycles. The red dashed lines indicate confidence bands for $1\sigma$ deviation. Right: corresponding plots for the mean of four measurements with reduced buffer gas pressure below $0.5\text{torr}$.

Figure 3

(Color online) Coherent and incoherent contributions to the Stokes signal for various write detunings. The three upper plots show frequency scans of the Stokes channel, using a Fabry-Perot resonator with a linewidth of $480\mathrm{MHz}$ and an FSR of $5\mathrm{GHz}$. The solid line is a fitted model, including a sharp Stokes peak (dotted) and a broader fluorescence peak (dashed) convoluted with the Fabry-Perot transmission function. The width of the fluorescence is assumed to be given by the Doppler profile according to the atomic temperature. The inset displays the approximate equality of peak separation and write detuning.