From Photons to Phonons and Back: A THz Optical Memory in Diamond

Optical quantum memories are vital for the scalability of future quantum technologies, enabling long-distance secure communication and local synchronization of quantum components. We demonstrate a THz-bandwidth memory for light using the optical phonon modes of a room temperature diamond. This large bandwidth makes the memory compatible with down-conversion-type photon sources. We demonstrate that four-wave mixing noise in this system is suppressed by material dispersion. The resulting noise floor is just $7\times {10}^{-3}$ photons per pulse, which establishes that the memory is capable of storing single quanta. We investigate the principle sources of noise in this system and demonstrate that high material dispersion can be used to suppress four-wave mixing noise in $\Lambda$-type systems.

Figure 1

(a) A Ti:sapphire oscillator ($\mathrm{Ti}:\mathrm{sa}$) at 800 nm is used to synchronously pump an OPO which creates the signal field at 723 nm. The remainder of the Ti:sapphire power provides the orthogonally polarized read-write pulses which are split, time delayed, and recombined on a polarizing beam splitter (PBS). Signal and read-write beams are overlapped on a dichroic mirror (DM) and focused into a diamond. After the diamond, interference filters (IF) remove the read-write beams and a polarizer (POL) selects the vertically polarized output mode which is detected on a photodiode (PD). (b) The memory states. The ground state $|0⟩$ is coupled to the excited optical phonon band $|1⟩$ by the signal and read-write beams which are in two-photon resonance with the phonon energy (40 THz). Both beams are far detuned from the conduction band $|2⟩$. (c) The 4WM mechanism. Two pump photons $P1$ and $P2$ are scattered to produce Stokes ($S$) and anti-Stokes ($AS$) photons, respectively, 19.

Figure 2

(a) The total memory efficiency as a function of read-write delay (10 nJ each in read-write pulses). An exponential fit of 3.5 ps represents decay of the optical phonons into a pair of acoustic phonons. (b),(c) Total efficiency (at 1 ps storage time) and write efficiency, respectively, as a function of read-write pulse energy. The power dependence of the memory efficiencies is fitted by solving the system in Eq. (1).

Figure 3

The degree of first-order coherence between the memory output and a reference pulse (red crosses) remains high despite decaying memory efficiency (black dots). This confirms that the memory coherence is unaffected by decay of the optical phonon.

Figure 4

(a) Memory output (black dots) with 0.6 photons per pulse at the input. With the input field blocked, the noise output is shown (red crosses) demonstrating good signal-to-noise contrast. The noise data are fitted with the phenomenological model outlined in Eq. (2) (thin solid red line). Thick transparent lines indicate the signal (black) and noise (red) measurements predicted by solving the coupled equations in Eq. (1). (b) Phase-matching diagram of the memory interaction showing perfect phase matching. (c) 4WM phase-matching diagram: A phase mismatch $\delta K$ arises from the material dispersion and suppresses the 4WM process.