Room-temperature high-speed nuclear-spin quantum memory in diamond

Quantum memories provide intermediate storage of quantum information until it is needed for the next step of a quantum algorithm or a quantum communication process. Relevant figures of merit are therefore the fidelity with which the information can be written and retrieved, the storage time, and also the speed of the read-write process. Here, we present experimental data on a quantum memory consisting of a single ${}^{13}$C nuclear spin that is strongly coupled to the electron spin of a nitrogen-vacancy (NV) center in diamond. The strong hyperfine interaction of the nearest-neighbor carbon results in transfer times of 300 ns between the register qubit and the memory qubit, with an overall fidelity of 88$\%$ for the write-storage-read cycle. The observed storage times of 3.3 ms appear to be limited by the $T_1$ relaxation of the electron spin. We discuss a possible scheme that may extend the storage time beyond this limit.

Figure 1

(a) Structure of the NV center with one ${}^{13}$C nuclear spin in the first shell. (b) Energy levels and transitions of the subsystem spanned by the $m_S=0$ and $m_S=1$ states of the electron spin and the two ${}^{13}$C nuclear spin states. (c) Pulse sequence for measuring the ${}^{13}$C Rabi oscillation and the resulting signal. (d) Pulse sequence for measuring the ${}^{13}$C FID and the resulting signal.

Figure 2

Initialization scheme and resulting fidelity. (a) Pulse sequence used. Since the states $|1,\downarrow \rangle$ and $|1,\uparrow \rangle$ are initially both unpopulated, one MW${}_1$ and one RF${}_1$ $\pi$ pulse are sufficient to generate a swap operation. The laser pulse polarizes the electron spin and partly depolarizes the nuclear spin. (b) Rate equation model for analyzing the effect of the laser illumination on the populations. (c) Experimental (points) and calculated (curves) populations as a function of the laser pulse duration. (d) Rabi oscillations with and without ten cycles of the initialization sequence. With (without) the initialization sequence, the fidelity is 80$\%$ (50$\%$).

Figure 3

Coherence transfer from NV electron spin to ${}^{13}$C nuclear spin. Upper part: Pulse sequence for coherence transfer and subsequent free evolution of ${}^{13}$C. The phase $\varphi$ of the initial $\pi /2$ pulse determines the phase of the superposition state, which is then transferred to the ${}^{13}$C quantum memory. Lower part: Free evolution of ${}^{13}$C. The initial phases of the Ramsey fringes reflects the phase $\varphi$ of the initial $\pi /2$ microwave pulse.

Figure 4

Survival of ${}^{13}$C nuclear spin coherence in spin-echo sequences. The experimental data points are compared to the electron spin-lattice relaxation (black curve).

Figure 5

Pulse sequence for a dynamical decoupling scheme that could extend the storage time beyond the $T_1^e$ limit.

Figure 6

Upper part: The measured rates in our rate-equation model of the populations [$\alpha$ (blue), $\beta$ (purple), and $\gamma$ (green)] as a function of laser power. Lower part: Calculated population $P_{|0,\uparrow \rangle }$ after the four repetitions of the initialization sequence.