A scalable gallium-phosphide-on-diamond spin-photon interface

Abstract: The efficient interfacing of quantum emitters and photons is fundamental to quantum networking. Quantum defects embedded in integrated nanophotonic circuits are promising for such applications due to the deterministic light-matter interactions of high-cooperativity ($C>1$) cavity quantum electrodynamics and potential for scalable integration with active photonic processing. Silicon-vacancy (SiV) centers embedded in diamond nanophotonic cavities are a leading approach due to their excellent optical and spin coherence, however their long-term scalability is limited by the diamond itself, as its suspended geometry and weak nonlinearity necessitates coupling to a second processing chip. Here we realize the first high-cooperativity coupling of quantum defects to hybrid-integrated nanophotonics in a scalable, planar platform. We integrate more than 600 gallium phosphide (GaP) nanophotonic cavities on a diamond substrate with near-surface SiV centers. We examine a particular device with two strongly coupled SiV centers in detail, confirming above-unity cooperativity via multiple independent measurements. Application of an external magnetic field via a permanent magnet enables optical resolution of the SiV spin transitions from which we determine a spin-relaxation time $T_1>0.4$ ms at 4 K. We utilize the high cooperativity coupling to observe spin-dependent transmission switching and the quantum jumps of the SiV spin via single-shot readout. These results, coupled with GaP's strong nonlinear properties, establish GaP-on-diamond as a scalable planar platform for quantum network applications.

• The paper demonstrates a scalable GaP-on-diamond platform enabling deterministic SiV coupling to photonic cavities, with simulated Q factors up to 1.75×10^5 and single-shot readout fidelity exceeding 96%.
• It employs a hybrid forward and inverse design strategy to optimize the cavity’s Q/V and cooperativity while systematically quantifying intrinsic, extrinsic, and fabrication-induced losses.
• The study integrates high-yield nanofabrication, precise stamping, and spectroscopic techniques to achieve robust quantum spin initialization, with T1 exceeding 400 μs and efficient fiber-to-chip coupling.

Scalable GaP-on-Diamond Spin-Photon Interface: Nanofabrication, Integration, and Quantum Readout

Device Design and Cavity Optimization

The paper presents a fully integrated GaP-on-diamond nanophotonic platform targeting scalable spin-photon interfaces with deterministic single-defect (SiV center) coupling for quantum information applications. The core device is a GaP photonic crystal (PhC) nanobeam cavity, optimized for strong evanescent coupling to silicon-vacancy (SiV) centers implanted shallowly into the underlying diamond substrate.

Key elements of the cavity optimization include:

• Quadratic chirping of the lattice constant to localize the optical mode (minimizing $V$ and maximizing $Q$) while preserving significant overlap with SiV dipoles positioned ≈20 nm below the interface.
• Hybrid design combining forward-design (maximizing $Q/V$) and inverse design via global (LIPO) and local (trust-region) search methods to directly maximize the expected cooperativity $C$, which encapsulates the Purcell enhancement and spatial overlap with the emitter.
• Separation of loss contributions into intrinsic radiative loss, extrinsic waveguide-coupling, and fabrication-induced losses, allowing a systematic quantification of experimentally achievable performance.

Evaluation of the objective function for the optimized design yields simulated $Q_i = 1.75 \times 10^5$, $V = 1.86(\lambda/n)^3$, and field overlap $|E_y(\mathbf{x}_a)/\max|\mathbf{E}| |^2 = 0.25$. For a realistic fabrication-constrained $Q_{\rm fab}=5 \times10^4$, the lower-bound cooperativity is $C=23$ (vs. $C=102$ for ideal fabrication), confirming strong potential for the system to operate in high-cooperativity (and thus high-fidelity spin-photon interface) regimes.

Figure 1: Nanobeam cavity design: (A) device cross-section with parameter labels; (B) optimized lattice constant profile; (C) simulated band structures illustrating midgap resonance and photonic bandgap contrast.

Fabrication, Stamping, and Integration Protocol

The integration platform exploits a heterogeneously fabricated GaP nanophotonic cavity that is transferred onto a planar diamond chip hosting a shallow ensemble of SiV centers. The process entails:

• EBL patterning of etched GaP-on-AlGaP membranes to define the PhC cavity arrays and embedding frames for mechanical support throughout the process.
• Controlled undercut and critical-point drying to release free-standing nanodevices.
• Polymer stamp transfer using a PC/PDMS composite, leveraging temperature-driven PC melt to pick and deterministically align hundreds of devices onto defined regions of the cleaned diamond chip.

Post-stamping, the PC film is dissolved, leaving the nanophotonic devices atop the SiV-rich diamond surface with >99% yield.

Figure 2: Overview of the GaP nanobeam device fabrication flow, from EBL and etch through release and final transfer to diamond.

Figure 3: Stamping sequence: sequential optical micrographs document device pickup, alignment, transfer, and release onto the diamond host.

Nanophotonic Coupling and Excitation

Efficient fiber-to-chip and chip-to-free-space coupling is critical for high-throughput device metrology and scalable photonic interconnects.

• Elliptical free-space grating couplers are co-designed with the PhC cavities, providing broadband transmission (≈20%) into a narrow angular profile, facilitating both excitation and collection via high-NA objectives.
• SEM and FDTD characterization validate the uniformity and spectral fidelity of the gratings. The primary bottleneck in system efficiency is traced to the grating’s numerical aperture-collection mismatch.

  Figure 4: (A) SEM of elliptical grating; (B) FDTD-modeled broadband TE/TM reflection and transmission showing dominant TE-mode coupling.

  

  Figure 5: Schematic of the integrated cryogenic microscope: laser, DAQ, and detector submodules enabling high-resolution, resonant, and off-resonant measurements.

Spectroscopic Characterization and Gas Tuning

To match device resonances to specific SiV transitions, xenon gas condensation is used to controllably redshift cavity modes, achieving reliable spectral overlap.

Figure 6: Resonance tuning and $Q$ versus xenon load: maximal shift is observed when PC holes fill, but is limited by cryostat geometry.

Transmission spectra are acquired both in broadband (supercontinuum) and narrowband (scanning diode laser, with high-precision wavemeters) regimes. Lorentzian fits to broadband spectra extract cavity $Q$ and spectral homogeneity metrics, while narrowband scans precisely resolve emitter-cavity interactions and allow evaluation of dipole-induced transparency (DIT).

Figure 7: Example broadband transmission measurement: Lorentzian fit isolates TE cavity response atop TM and background contributions.

Figure 8: Multiple high-resolution DIT scans show dual high-contrast dips for SiVs coupled to a single cavity, enabling direct extraction of coupling rates.

Experimental Results: High-Cooperativity Spin-Photon Coupling

Multiple device instances are characterized for quantum emitter-cavity coupling:

• DIT features are robustly observed in both “drop” and “thru” geometries, with contrasts dependent on design and coupling configuration.
• Quality factors of $Q \sim 3000 - 10000$ are measured post-integration.
• Lower bounds on single-emitter cooperativity range from $C > 0.36 \pm 0.01$ to $C > 0.50 \pm 0.02$, with all values suppressed by residual spin and orbital thermalization. These values approach or surpass state-of-the-art for hybrid photonic/dipole platforms, confirming strong-coupling regime accessibility.

  Figure 9: Transmission (broadband and DIT) for a drop-coupled device with $Q=4300$; lower-bound $C>0.50$ from input-output model fit.

  

  Figure 10: Single-peak DIT for drop-coupled device with $Q=2900$ and $C>0.36$.

  

  Figure 11: Thru-coupled device: $Q=4200$, two resolved DIT dips; fit constraints provide $C>0.45$ lower bound.

Spin Initialisation, Relaxation, and Optical Quantum Readout

A custom magnetic mount aligns the external field to selected SiV principal axes, stabilizing the spin basis for optical pumping and readout. Fine-tuning of field angle and strength is corroborated by modeling and experimental transition splittings.

• Spin relaxation $T_1$ is quantified in both misaligned ($\alpha \sim 70^\circ$) and near-aligned ($\alpha \lesssim 4^\circ$) field configurations. In the optimal geometry, $T_1$ exceeds 400 μs—a value limited predominantly by magnetic field alignment and local defect environment, and in agreement with prior reports on SiV spin-lattice relaxation.

  Figure 12: 3D simulation and experimental schematics for the embedded SmCo magnetic mount and field-control protocols.

  

  Figure 13: Off-axis field experimental signatures: combined PLE and DIT, selective ionization, optical pumping and spin-recovery kinetics.

  

  Figure 14: Optical switching via resonant spin pumping and DIT recovery readout decays.

  

  Figure 15: Spin-pumping and $T_1$ decay for near-aligned field; robust two-level thermalization behavior.

Single-shot measurements are demonstrated, extracting spin-projected quantum-jumps in transmission, resulting in a single-shot readout fidelity $F = 96.0 \pm 0.3$\% for optimal thresholds, and confirming quantum non-demolition readout capability in the GaP-on-diamond photonic platform.

Figure 16: Raw time-resolved single-shot transmission, revealing quantum jumps, photon shot-noise, and blinking events.

Figure 17: Post-selected sequences: clear thresholding allows for >96% single-shot spin state discrimination.

Implications, Comparisons, and Outlook

By integrating deterministic, high-yield GaP nanocavity arrays directly onto high-purity diamond chips, this work achieves a substantial advance in scalable, hybrid quantum photonics. The achieved cooperativities and spin $T_1$ reservoirs closely match, or exceed, leading integrated quantum node demonstrations across SiV and related group IV color centers [Sipahigil et al., Science 2016; Evans et al., Science 2018]. The methodology's high device yield, field-insensitive fabrication, and post-transfer tuning pipelines support wafer-scale scaling, critical for multiplexed quantum repeater networks and distributed quantum memory arrays.

Several open avenues are noted:

• Further $Q$ enhancement: leveraging state-of-the-art lithographic and etch optimization, as well as surface passivation, should bridge the remaining gap to the $Q_i \gtrsim 10^5$ regime required for error-corrected network nodes and entanglement distribution [Bhaskar et al., Nature 2020; Stas et al., Science 2022].
• Full integration: coupling to waveguide buses and on-chip switching elements is compatible with the presented architecture, informed by recent advances in CMOS backend hybrid photonic integration [Li et al., Nature 2024], telecommunication-wavelength quantum networking [Bersin et al., PRX Quantum 2024], and hybrid quantum interconnects [Riedel et al., (Riedel et al., 8 Aug 2025)].
• Spin-photon interface expansion: the method is generalizable to alternative emitters (SnV, NV, rare-earth doped) and other wide-bandgap photonic platforms, as well as to composite functionalities such as non-reciprocity, acousto-optic tuning, and parametric processes enabled by the nonlinear properties of GaP [Wilson et al., Nat. Photonics 2020].

Conclusion

This work establishes gallium-phosphide-on-diamond nanophotonic cavities as a powerful, deterministic, and scalable spin-photon interface platform. The approach combines optimal cavity performance, robust fabrication and transfer, efficient optical access, and direct quantum state readout. These results define a high-cooperativity, high-fidelity standard for hybrid quantum node integration, bridging fundamental color center quantum optics, scalable photonic engineering, and quantum network deployments (2601.04733).