Electrically tunable, two-photon interference from remote silicon-vacancy centers in industrial silicon carbide

Technical Summary: Electrically Tunable, Two-Photon Interference from Remote Silicon-Vacancy Centers in Industrial Silicon Carbide

Problem Statement
The realization of scalable distributed quantum networks requires spatially separated network nodes that can emit indistinguishable photons with high spectral stability over prolonged timescales. While color centers in solid-state hosts (such as diamond) have demonstrated promising spin-photon interfaces, they face two primary challenges:

1. Spectral Inhomogeneity and Broadening: Independently created defects typically exhibit broad optical transition lines due to local charge noise and substantial inhomogeneous frequency distributions.
2. Long-Term Stability: Achieving and maintaining spectral matching between independent nodes is difficult. Short-term stability must be below the optical linewidth to enable interference, while long-term stability is required to minimize the overhead of re-calibration and tuning, thereby improving the network duty cycle.

Existing solutions often rely on pre-selecting specific centers within narrow frequency ranges or using complex local Stark-shift electrodes, which can be cumbersome for mass deployment.

Methodology
The authors address these challenges by utilizing silicon vacancy (VSi) color centers in 4H-silicon carbide (4H-SiC), a mature industrial semiconductor material. The core methodology involves:

• Device Integration: VSi centers were embedded in industrially available, laterally doped p-i-n diode structures fabricated on 4H-SiC wafers. The intrinsic region of the diode is 4.1 µm wide.
• Electrical Control: By applying a reverse bias voltage across the p-i-n diode, the authors achieve two complementary effects:
  • DC Stark Shift: Tuning the emission frequency of the VSi centers.
  • Charge Depletion: Suppressing fluctuations in the local electrostatic environment, which narrows the optical linewidth.
• Experimental Setup: Two separate VSi centers (VSi,1A and VSi,2B) were selected from different industrial-grade samples. Each sample was housed in an independent closed-cycle cryostat, separated by a distance of two meters.
• Measurement Protocol:
  • Characterization: 19 VSi centers were characterized to verify spectral tunability and linewidth narrowing.
  • Two-Photon Interference: A Hong-Ou-Mandel (HOM) experiment was conducted using off-resonant excitation (728 nm) to drive the A2 transition. The emitted photons were filtered via individual Fabry-Pérot cavities to isolate the A2 zero-phonon line (ZPL) before interfering on a free-space 50:50 beam splitter.
  • Stabilization: A dual-loop control system was implemented for a 628-hour (26-day) measurement campaign. A fast loop (10 s intervals) stabilized Fabry-Pérot cavity resonance, while a slow loop (2-hour intervals) monitored the absolute emission frequency via resonant photoluminescence excitation (PLE) and adjusted the diode bias to maintain spectral overlap.

Key Contributions

• Demonstration of Mass-Deployable Tunability: The work proves that VSi centers in standard industrial SiC wafers can be electrically tuned to a common target frequency without pre-selection.
• Spectral Narrowing: The p-i-n diode structure successfully narrowed the optical linewidths of the centers below 60 MHz across all 19 characterized samples, approaching the lifetime limit.
• Long-Term Remote Interference: The authors performed the first comprehensive investigation of two-photon interference between VSi centers in separate 4H-SiC samples held in separate cryostats.
• Automated Stability: The implementation of an automated feedback system allowed for sustained spectral matching over 26 days, requiring bias readjustment only every 8.4 hours on average for the less stable node.

Results

• Linewidths and Tuning: The 19 characterized centers showed linewidths below 60 MHz. The emission frequency exhibited a linear dependence on reverse bias (approx. 4.33 GHz/MV/m), allowing all centers to be tuned to a common target frequency ($\nu_T = 327.23100$ THz).
• Two-Photon Interference Visibility: The experiment achieved a raw two-photon interference visibility of 82.0% (0.820 ± 0.006).
  • The parallel polarization configuration yielded a normalized cross-correlation at zero delay of $G^{(2)}_{12,\parallel}(0) = 0.091(3)$, well below the classical limit of 0.5.
  • The orthogonal polarization control yielded $G^{(2)}_{12,\perp}(0) = 0.506$, confirming the distinguishability of the photons in that configuration.
• Stability Analysis:
  • The mutual detuning between the two centers had a mean of 1.9 MHz and a standard deviation of 15.5 MHz over the campaign.
  • The visibility gap (the difference between the measured 82% and the theoretical maximum) was quantitatively attributed to residual spectral diffusion (15.5 MHz standard deviation) and the non-Fourier-limited nature of the emitters, rather than experimental imperfections.
  • The system required bias correction every 19.4 hours for Sample A and 8.4 hours for Sample B, demonstrating significantly improved long-term stability compared to typical solid-state systems.

Significance
The paper establishes that silicon vacancy centers in industrial-grade 4H-SiC, when integrated into electrically controlled p-i-n diode structures, constitute a scalable and spectrally stable building block for distributed quantum networks. The results demonstrate that high two-photon interference visibility can be sustained over practical timescales (weeks) without the need for complex pre-selection or cryogenic nanophotonic structures. The 82% visibility aligns with the current state-of-the-art for remote solid-state emitters (matching results for remote tin-vacancy centers in diamond) while offering the advantage of using standard industrial semiconductor fabrication processes. This work validates the feasibility of using VSi centers in SiC for the realization of large-scale, multi-node quantum networks.