Efficient Quantum Memory Using a Weakly Absorbing Sample

A light-storage experiment with a total (storage and retrieval) efficiency $\eta =56\%$ is carried out by enclosing a sample, with a single-pass absorption of 10%, in an impedance-matched cavity. The experiment is carried out using the atomic frequency comb (AFC) technique in a praseodymium-doped crystal ($0.05\%{\mathrm{Pr}}^{3+}:{\mathrm{Y}}_2\mathrm{Si}{\mathrm{O}}_5$) and the cavity is created by depositing reflection coatings directly onto the crystal surfaces. The AFC technique has previously by far demonstrated the highest multimode capacity of all quantum memory concepts tested experimentally. We claim that the present work shows that it is realistic to create efficient, on-demand, long storage time AFC memories.

Figure 1

(a) Part of the experimental setup. A frequency stabilized ($<1\mathrm{kHz}$ linewidth) dye laser at ${\lambda }_{\mathrm{vac}}=605.977\mathrm{nm}$ is employed as light source. A $50/50$ beam splitter splits off part of the input light onto a photo diode (PD1), to calibrate against variations in the input light intensity. Two other photo diodes monitor the transmitted (PD2) and the reflected (PD3) light from the cavity. The cavity length along the $b$ axis is $L=2\mathrm{mm}$ and the crystal diameter in the ($D_1$, $D_2$) plane is 12 mm. $b$, $D_1$, and $D_2$ are principal axes of the crystal [46]. A beam waist of about $100\mu \mathrm{m}$ is created at the crystal center via a lens with $f=400\mathrm{mm}$. A half-wave plate ($\lambda /2$) is employed to align the polarization direction to a principal axis of the cavity crystal, which is immersed in liquid helium at 2.1 K. The input and output facet of the crystal has $R1=80\%$ and $R2=99.7\%$ reflectivity. A small part of the cavity crystal is left uncoated for measurements without a cavity effect. (b) The hyperfine splitting of the ground $|g⟩$ and excited $|e⟩$ state of the ${}^{3}H_4{-}^1{D}_2$ transition of site I in ${\mathrm{Pr}}^{3+}:{\mathrm{Y}}_2\mathrm{Si}{\mathrm{O}}_5$ [38].

Figure 2

The normalized reflected signal (PD3/PD1) is plotted against the frequency offset from the ${\mathrm{Pr}}^{3+}:{\mathrm{Y}}_2\mathrm{Si}{\mathrm{O}}_5$ inhomogeneous line center. The crystal was translated perpendicular to the input beam (see Fig. 1) and the graph is a frequency scan for the position that gave the best impedance matching. At the impedance-matching condition the reflected light detected at PD3 should vanish. The best impedance-matched condition without memory preparation (i.e., spectral manipulation by the absorption profile, see text) was a reflection of 16% (84% absorption), which was measured about 45 GHz above the inhomogeneous broadening center frequency that is located at 0 GHz in this figure.

Figure 3

The absorption profile created through the optical pumping process is shown. This measurement is done in a small part of the cavity crystal where the coating was removed. From the Pr inhomogeneous absorption profile, four AFC peaks, all absorbing on the $|\pm \frac{5}{2}g⟩\mapsto |\pm \frac{5}{2}e⟩$ transition, are created. The spectral content of the storage pulse is shown schematically in the inset by the red dashed line across the AFC structure, which has peak width $\gamma$ and peak separation $\Delta$. The finesse of the AFC structure is $F_{\mathrm{AFC}}=\frac{\Delta }{\gamma }$.

Figure 4

The input storage pulse as a reference detected at PD3 (see text) is shown as a red dashed line. The retrieved echo pulse is detected at detector PD3 after $1.1\mu \mathrm{s}$ as shown with the black solid line. The area of the echo pulse at $1.1\mu \mathrm{s}$ divided by the reference signal pulse area gives a storage and retrieval efficiency of the memory of $\eta =56\%$.