Chiral microwave metasurface for controlling spins in diamond
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | npj Nanophotonics |
| Published | 2026-05-29 |
| Authors | Kosuke Takada, Ryota Katsumi, Shusuke Akagi, Kazunobu Hashimoto, Takashi Yatsui |
| DOI | 10.1038/s44310-026-00138-0 |
A glide-symmetric spoof surface plasmon polariton metasurface produced a near-circular microwave magnetic field with a local degree of circularity D ≈ 1 near the patterned surface. ODMR measurements showed direction-dependent selection of NV spin transitions and magnetic spin-momentum locking. The simulated field could also align with different NV axes at separate positions, which supports the authors’ proposal for on-chip zero-field vector magnetometry.
A single-mode alternative to phase control
Section titled “A single-mode alternative to phase control”Conventional circularly polarized microwave magnetic fields are usually generated by superposing fields from two crossed or parallel transmission lines. This approach requires precise control of the phase difference between the lines. Transmission lines also confine the microwave field weakly because their dispersion lies near the light line.
Microwave cavities can confine the field temporarily, but a higher quality factor limits bandwidth. Their operation also requires phase-controlled excitation through multiple ports. The SSPP approach instead produces the polarization within one propagating mode. The authors state that phase control is unnecessary for this mode. Its two-dimensional geometry is compatible with printed circuit boards and semiconductor chips.
Geometry and confinement at the NV transition
Section titled “Geometry and confinement at the NV transition”The eigenmode calculation used groove height h = 18 mm, groove width a = 1 mm, unit-cell length b = 4 mm, and center waveguide width w = 1 mm. COMSOL Multiphysics computed Bloch modes for the periodic x direction. The model used a perfect electric conductor pattern on 0.508 mm Rogers RT/Duroid 5880 with a perfect electric conductor base.
The calculated lowest-order mode followed the dispersion relation of a surface plasmon polariton. The NV center zero-field splitting of about 2.88 GHz lay near the mode’s asymptotic region. The authors associated this region with strong microwave field confinement. They evaluated the polarization at $k_x b/\pi = 0.9$ and at heights of z = 0, 0.5, 1, and 1.5 mm above the metal pattern.
The polarization texture changes with height
Section titled “The polarization texture changes with height”Near the patterned surface, the photonic spin vector derived from the magnetic field pointed normal to the surface. Its direction reversed after every half unit cell. At the intersection of the center and side waveguides, the calculated degree of circular polarization reached D ≈ 1. This near-surface field projects equally onto all NV axes in the analyzed (100)-oriented diamond structure.
As the observation height increased, the polarization vectors tilted away from the surface normal. At z = 1.5 mm, the simulated orientation could align with the four possible NV axis directions. The degree of circularity remained spatially nonuniform because electromagnetic coupling between unit cells distorted its pattern. The authors therefore used both the polarization direction vector and degree of circularity to map the field.
The fabricated microwave and diamond device
Section titled “The fabricated microwave and diamond device”The experimental pattern used the same h, a, and b values, with w = 0.5 mm. Copper layers were 35 μm thick on both sides of the 0.508 mm Rogers substrate. Laser processing formed the SSPP pattern. Each CPW-SSPP taper combined gradient grooves with an exponentially curved ground pattern. The gradient-groove and curved-ground parts each used 15 unit cells.
Diamond microrings containing an ensemble of NV centers were made by angled etching. The source was a DNV-B1 diamond sample from Element Six with an NV density of approximately 300 ppb. Transfer printing placed the microrings on a 20 × 20 × 0.05 mm cover glass. Adhesive tape attached the cover glass to the center of the microwave device. Because the cover glass was approximately 50 μm thick, the photonic spin vector remained aligned with the diamond [100] direction over the glass.
This configuration gave equal projections of the circular field onto the transverse plane of the [111], [1̄1̄1], [1̄11̄], and [11̄1̄] NV axes. The paper states that this geometry supports spin-selective excitation of the four orientations in the weak-power regime.
ODMR measurement and spin-momentum locking
Section titled “ODMR measurement and spin-momentum locking”The free-space ODMR setup used continuous-wave 532 nm laser light with a 0.8 numerical aperture objective. Laser power at the objective was 2 mW. An avalanche photodiode collected NV fluorescence, while an RFSoC and the QICK-DAWG platform controlled microwave excitation and readout. The signal and reference windows were 200 μs, and each frequency was measured 5,000 times. A bias magnetic field separated the ODMR dips during this experiment.
The microwave chain used two bandpass filters and a 45 dB amplifier. The fed power was -9 dBm, which kept the measurement in a low-spin-excitation regime. This condition suppressed linewidth broadening and allowed the recognized hyperfine splitting. Scattering measurements showed low-pass behavior because the structure did not support a propagating mode at higher frequencies. No resonant standing-wave pattern appeared in transmission.
For sufficiently weak microwave power, the maximum transition probabilities could reach (2 + √3)/12 and (2 − √3)/12 for the $m_s = 0$ to $m_s = ±1$ transitions. The circular field therefore addressed one spin branch more strongly than the other. This selection predicts the reduction from eight conspicuous ODMR dips to four.
The measured spectra showed strong selection of NV transitions for propagation along +x. Propagation along -x selected transitions associated with the opposite circular polarization and suppressed the first set. The authors describe this as a high extinction ratio, but give no numerical value. This behavior is the observed magnetic spin-momentum locking. It differs from earlier SSPP demonstrations of electric spin-momentum locking and conventional electric-based chiral optics.
Alignment for zero-field vector magnetometry
Section titled “Alignment for zero-field vector magnetometry”Frequency-domain simulations tested three-dimensional alignment at 2.88 GHz with w = 0.5 mm. On the z = 1.5 mm plane, |cos θ1| was approximately 1 within every unit cell. At the points aligned with the [111] NV axis, the degree of circularity satisfied 0.8 < D ≲ 1. The same line contained alignment points for the other NV axes. Each axis could therefore be addressed at a different position without rotating the diamond.
The authors discuss integrating NV-containing nanostructures into areas on the scale of 10 μm with |θm| < 3°. They propose using the spatial alignment for on-chip zero-field vector magnetometry. The full axis-selective, three-dimensional alignment remained a simulation result.
Limits and engineering implications
Section titled “Limits and engineering implications”Several limits remain. The measured ODMR spectra used an external bias field to separate the resonances, so the proposed zero-field mode was not tested directly. The paper reports no numerical extinction ratio, magnetometric sensitivity, conversion efficiency, or bandwidth. The article is also an unedited version, and the authors warn that errors may affect the content. The datasets are available from the corresponding author on reasonable request because their size and raw data prevent public release.
Diamond device engineering therefore has to account for the NV ensemble, microring position, and three-dimensional field map. The design maps spin transitions to specific field positions, removing the need to rotate the diamond when selecting among NV axes. The paper’s engineering result is a passive mode that supplies confined, circular, nonreciprocal microwave excitation.
Materials and support
Section titled “Materials and support”6CCVD’s diamond materials catalog lists MPCVD-grown single-crystal diamond plates and wafers up to 125 mm, with thickness from 0.1 to 500 μm and Cu metallization. These items cover the base material and the paper’s 35 μm copper layers for its microwave spin-control device, but not its approximately 300 ppb NV ensemble or angled-etched microring geometry. Engineering support is the appropriate route for questions about those device-specific integration requirements.
View Original Abstract
Abstract Nitrogen-vacancy (NV) centers in diamond are versatile platforms for quantum technologies, for which engineering microwave magnetic-field confinement and polarization are key elements. However, current microwave devices lack the capability to flexibly manipulate both. In this study, we address this issue by investigating a spoof surface plasmon polariton (SSPP)-based metasurface for engineering strongly confined microwave magnetic fields. Within the propagating SSPP mode, we reveal a rich texture of three-dimensionally oriented polarization, including circular polarization. In addition, we demonstrate the generation of a circularly polarized microwave magnetic field for an NV-containing diamond nanostructure, evidenced by spin-resonance spectra probed via the SSPP. We further confirm the non-reciprocal behavior of spin excitation due to the magnetic spin-momentum locking of propagating SSPPs. Furthermore, the circularly polarized field can be engineered to align with NV axes, enabling spin- and axis-selective control. This architecture could potentially enable on-chip zero-field vector magnetometry, overcoming a significant hurdle in compact quantum device integration.
Electronic Grade Intrinsic Diamond Epitaxial Wafer H Terminated Superior 100 Mosfet Ready Substrate
View Specs →Products
View Specs →