Two-Photon Quantum Interference from Separate Nitrogen Vacancy Centers in Diamond

We report on the observation of quantum interference of the emission from two separate nitrogen vacancy (NV) centers in diamond. Taking advantage of optically induced spin polarization in combination with polarization filtering, we isolate a single transition within the zero-phonon line of the nonresonantly excited NV centers. The time-resolved two-photon interference contrast of this filtered emission reaches 66%. Furthermore, we observe quantum interference from dissimilar NV centers tuned into resonance through the dc Stark effect. These results pave the way towards measurement-based entanglement between remote NV centers and the realization of quantum networks with solid-state spins.

Figure 1

(a) Optical microscope image of the sample. The microwave stripline (MW) is used for spin manipulation and the gates (G1 and G2) are used for Stark tuning of the optical transition energies. Inset: Scanning electron microscope image of a similar device. (b) Confocal microscope image showing NV $A$ and NV $B$ (logarithmic color scale). (c) Experimental setup. The two NV centers inside the same diamond are simultaneously excited with either a resonant (637 nm) or an nonresonant (532 nm) laser. Two separate paths allow detection of photons emitted into the zero-phonon line (ZPL) or the phonon side band (PSB). A variable retardance wave plate (WP) aligns the polarization of the desired transition to the polarizing beam splitter (PBS) while compensating for ellipticity introduced by other optical components.

Figure 2

Spectral characterization of NV $A$ and NV $B$. (a) Level structure. The $m_s=0$ ground state is connected via spin selective transitions to the $E_x$ and $E_y$ ($m_s=0$) excited state, $m_s=\pm 1$ states are omitted for clarity. Photons that are emitted into the phonon side band (PSB) can be spectrally separated from resonant emission. Also shown are the resonant (637 nm) and the nonresonant laser (532 nm). (b) Resonant excitation of the ZPL and detection of the PSB. The frequency of the red laser is swept across the resonances while weak green excitation prevents photoionization and optical pumping (see text). The spectra are averaged over 70 scans. (c) Scanning Fabry-Perot spectrum under nonresonant excitation. Only the $E_x$ transition is visible confirming sufficient suppression of the $E_y$ transition by polarization filtering. The linewidths (450 MHz for NV $A$ and 570 MHz for NV $B$) include drifts of the cavity during the measurement. The offset is due to detector dark counts.

Figure 3

Two-photon quantum interference. (a) Orthogonal polarization. Periodic peaks of the coincidence counts correspond to the 10 MHz repetition frequency of the excitation. The coincidence distribution shows typical bunching and antibunching features of the two independent NV centers under pulsed excitation [37]. (b) Parallel polarization. Interference of indistinguishable photons leads to a significant decrease in coincidence events at zero time delay. Simulations (solid line) using only independently measured parameters are done according to Legero et al. [27]. The dashed lines show 1 standard error uncertainty in the simulation. The dark and light grey areas illustrate 1 and 2 standard deviations expected scatter. Parameters used for the simulations are: excited state $\mathrm{\text{lifetime}}=12.0\pm 0.4\mathrm{ns}$, detuning between the two $\mathrm{\text{centers}}=130\pm 150\mathrm{MHz}$, frequency jitter between $\mathrm{\text{them}}=550\pm 80\mathrm{MHz}$, APD $\mathrm{\text{jitter}}=410\pm 10\mathrm{ps}$, dark count $\mathrm{\text{rate}}=60\pm 10{\mathrm{s}}^{-1}$, NV B count $\mathrm{\text{rate}}=1470\pm 50{\mathrm{s}}^{-1}$, NV $A$ count $\mathrm{\text{rate}}=2700\pm 50{\mathrm{s}}^{-1}$, $\mathrm{\text{background}}=15\pm 5\%$.

Figure 4

Interference of two dissimilar sources. (a) $E_y$ transition energy as a function of applied gate voltage. The two centers show opposite tuning behavior and are brought into resonance at $-13.6\mathrm{V}$. Data are taken under simultaneous green excitation with the same power that is used for the interference measurement. (b) Two-photon quantum interference. The overall coincidence distribution is the sum of 255 one-minute histograms. Before each interference measurement we perform a PLE scan. We select only histograms for which the relative detuning of the two centers is between 350 and 1200 MHz. The simulation (red line) is based on the measured frequency distribution and shows oscillations that reflect the discrete set of detuning values. The dark and light grey areas illustrate the expected signal for two noninterfering sources with 1 and 2 standard deviations, respectively.