Germanium Vacancy in Diamond Quantum Memory Exceeding 20 ms

Negatively charged group-IV defects in diamond show great potential as quantum network nodes due to their efficient spin-photon interface. However, reaching sufficiently long coherence times remains a challenge. In this work, we demonstrate coherent control of germanium vacancy center (GeV) at millikelvin temperatures and extend its coherence time by several orders of magnitude to more than 20 ms. We model the magnetic and amplitude noise as an Ornstein-Uhlenbeck process, reproducing the experimental results well. The utilized method paves the way to optimized coherence times of group-IV defects in various experimental conditions and their successful applications in quantum technologies.

Figure 1

(a) Reduced energy level scheme showing the lower orbital branches in the ground (green) and excited (blue) state manifold. The electron spin becomes accessible by applying an external magnetic field $B=100\mathrm{mT}$ (b) through the optical transitions $c_{1,2}$. (c) Corresponding PLE spectrum under constant MW repumping at frequency ${\nu }_0$. Slight misalignment of $\overrightarrow{B}$ to the GeV axis allows for efficient optical initialization of 98% within 1 ms (d).

Figure 2

(a) ODMR spectrum (shown in green) shows transition frequencies ${\nu }_{1,2}$ with 300 kHz linewidth. The separation by 2.98 MHz indicates a strongly coupled ${}^{13}\mathrm{C}$. For further measurement, the driving field ${\nu }_0=3.066\mathrm{GHz}$ was chosen such that both transitions are equally covered (shown in blue). (b) Corresponding Rabi oscillations with a frequency of $\mathrm{\Omega }=(2\pi )6.486\mathrm{MHz}$. (c) Ramsey interference measurement reveals $T_2^{*}=1.43\mathrm{\mu }\mathrm{s}$. (d) Hahn echo decay measurement yields spin-noise-limited $T_2=440\mathrm{\mu }\mathrm{s}$.

Figure 3

(a) CPMG sequence with $N$ refocusing $\pi$ pulses separated by $\tau$. The phase of each pulse is indicated in a subscript. (b) Decay curves with fixed $N=\{1,2,4,8\}$ and increasing $\tau$. Dashed lines show fits of the envelopes to $\exp [-(N\tau /T_2{)}^{\beta }]$ with $\beta$ a free parameter. Inset: Enlargement of $N=2$ measurement, showing regular dips due to entangling and disentangling to a ${}^{13}\mathrm{C}$.

Figure 4

(a) Memory measurement sequence using the CPMG (b) and $XY8$ (c) protocol. Subscripts represent the phase of a pulse. (d) Experimental results for CPMG (blue dots) and $XY8$ (green triangles) with fixed $\tau =100\mathrm{\mu }\mathrm{s}$ and sweeping order $N$, $\tilde{N}$ vs the total duration time $T$. Fluorescence is normalized to the expected value for $T=0$ for each sequence [23]. Exponential decay fitting yields $T_{2,\mathrm{CPMG}}=24.1\pm 0.9\mathrm{ms}$ and $T_{2,XY8}=18\pm 3\mathrm{ms}$. Solid lines represent OU simulations for CPMG (blue) and $XY8$ (green), closely matching the measured data. The results remain within the simulation curves for the boundaries of the correlation time of 12.4 s (18.6 s), displayed as dashed (dash-dotted) lines (see the text and [23]).