Magnonic cavity-enhanced quantum magnetometry with nitrogen-vacancy centre ensembles

The gist: This work proposes a chip-scale NV magnetometer built around a one-dimensional YIG/CoFeB magnonic crystal, in which a single structural defect forms a passive magnonic cavity and integrated permanent magnets supply the bias field.

How it works

1. The system utilizes a one-dimensional YIG/CoFeB magnonic crystal where a single structural defect creates a passive magnonic cavity within the band gap, and integrated permanent magnets provide the in-plane bias field H0 without needing external electromagnet hardware (Page 3).

2. Micromagnetic simulations identify a defect-localised spin-wave mode at 2.910 GHz inside the 2.5–3.0 GHz magnonic band gap, which matches an independent plane-wave calculation (Page 4).

3. The NV ensemble couples to this magnon mode, and a Tavis–Cummings/Lindblad model predicts a hyperfine-resolved optically detected magnetic resonance spectrum that is strong only within the mode linewidth, and a dispersive homodyne response with one steep edge per 14N hyperfine line (Page 3).

4. The theoretical framework involves a Tavis–Cummings-type Hamiltonian describing an NV ensemble coupled to a single bosonic mode, where the transverse dipolar field of the defect-localised mode sets the Rabi frequency Omega and the field per magnon sets gens (Page 4).

Magnonic Band Gap and Defect Mode

"Micromagnetic simulations with MuMax3 identify the defect-localised mode and its dynamic dipolar field at the NV site; a plane-wave method (PWM) independently verifies the band structure; and a Tavis–Cummings/Lindblad model solved in QuTiP converts this field into hyperfine-resolved spectra and multispectral sensitivity"

Inside this gap, a single narrow resonance appears at flocal = 2.910 GHz, close to the NV zero-field splitting fZFS = 2.87 GHz

The defect-localised mode is classified as genuine only if its frequency lies inside a gap of the pristine band structure rather than on a folded bulk band, and its spatial profile must be exponentially confined to the defect rather than extended across the crystal (Page 21).

Magnon-Dressed NV Spectra and Readout

The antenna does not drive the NV centres directly; it excites the defect mode, whose transverse dipolar field drives the ms = 0 ↔ ms = +1 transition of each hyperfine sector

A passive magnonic cavity therefore drives strongly only the hyperfine line that lies within its linewidth

The resulting ODMR spectrum shows three distinct lines at any single bias field, where a mode linewidth comparable to 2Ahf would drive all three lines nearly equally (Page 6). The homodyne readout provides one steep dispersive edge per hyperfine line, which is the magnetometer’s operating point (Page 8).

Sensitivity Projection and Design Levers

The projected three-edge sensitivity of 93 pT/√Hz is approximately eight times lower than the 12 pT/√Hz measured with the driven dielectric cavity (Page 8). The analysis identifies key design levers to close this gap, including the antenna–mode coupling, which should match the total loss of the mode including spin absorption, leading to an optimum antenna coupling of κ opt ext ≈ κint + 4g 2ens/Γ (Page 8).

The optimal NV–magnon coupling gens/2π is found near gens/2π ≈ 0.44 MHz for a spin linewidth of Γ/2π = 0.64 MHz, reaching 72 pT/√Hz (Page 8). Narrowing the spin linewidth to 0.17 MHz improves the optimum to 19 pT/√Hz near gens/2π ≈ 0.20 MHz (Page 8).

Validation and Conclusion

Both the ODMR route and the homodyne route show strong agreement between theory and simulation, with R2 > 0.9999 for the homodyne map (Page 8). The projected three-edge sensitivity of 93 pT/√Hz is achieved with a multiplexing gain of 1.72, which is close to the √3 expected for three edges of nearly equal slope (Page 8). This work shows that cavity-coupled sensing survives replacing an actively driven photon mode with a passive, defect-localised magnon mode (Page 9).

This result opens a design space termed magnon-mediated multispectral sensing, which complements and can be tested against its cavity-driven counterpart (Page 9). The analysis confirms that fabrication fixes the levers—antenna–mode coupling, NV–magnon coupling gens, spin linewidth—favoring size, weight, power, and thermal budget in chip-scale deployment (Page 11). This research provides a physically grounded simulation showing that cavity-coupled readout in cavity-driven NV magnetometry survives replacing an actively driven photon mode with a passive, defect-localised magnon mode (Page 9).

The natural next steps include multi-defect crystals, analogous to coupled-cavity arrays, self-consistent micromagnetic–QuTiP co-simulation including magnon–NV back-action, and experimental verification of the predicted localised mode in a fabricated crystal integrated with a diamond NV layer (Page 9). For MAD and passive geomagnetic navigation, a cavity-free, low-power sensor head is attractive even though MAD detection range grows only as the cube root of any sensitivity improvement (Page 9). The technical noise excess of 10.9 for the practical readout chain is applied to project the final sensitivities (Page 8). The sensitivity is constant up to about 100 kHz and rises above a −3 dB bandwidth of about 360 kHz, set by the spin and mode linewidths (Page 8). The response is flat while f is small compared with the linewidths of the coupled system, so the sensitivity is constant up to about 100 kHz (Page 8). This work shows that cavity-coupled sensing survives replacing an actively driven photon mode with a passive, defect-localised magnon mode (Page 9). The result opens a design space, termed here magnon-mediated multispectral sensing, that complements and can be tested against its cavity-driven counterpart (Page 9). Natural next steps include multi-defect crystals, analogous to coupled-cavity arrays, for independent multiplexed readout (Page 9). For MAD and passive geomagnetic navigation, a cavity-free, low-power sensor head is attractive even though MAD detection range grows only as the cube root of any sensitivity improvement (Page 9). The technical noise excess of 10.9 for the practical readout chain is applied to project the final sensitivities (Page 8). The projection assumes that the spins respond linearly at the probe power; this requires a sufficiently large ensemble and should be verified for a specific device (Page 8). The sensitivity is flat at 93 pT/√Hz up to about 100 kHz and rolls off above the 360 kHz bandwidth set by the spin and mode linewidths (Page 8). The small features near 2 MHz appear where the modulation sidebands reach its neighbouring hyperfine lines, separated by Ahf = 2.16 MHz (Page 8). This work shows that cavity-coupled sensing survives replacing an actively driven photon mode with a passive, defect-localised magnon mode (Page 9). The result opens a design space, termed here magnon-mediated multispectral sensing, that complements and can be tested against its cavity-driven counterpart (Page 9).

1. Improved sensitivity extraction via magnon-mediated readout: The system can achieve a projected sensitivity of 93 pT/√Hz by utilizing a passive, defect-engineered 1D YIG/CoFeB magnonic crystal, which is a chip-scale sensing element that avoids the limitations of actively pumped dielectric resonators.

2. Enhanced spectral resolution for magnetic field mapping: The system can provide hyperfine-resolved optically detected magnetic resonance spectrum[s] by coupling the NV ensemble to a defect-localised spin-wave mode, resulting in a dispersive homodyne response with one steep edge per 14N hyperfine line.

3. Optimized readout hardware design: The AI system can determine the optimal antenna coupling by calculating that a slightly over-coupled antenna is therefore preferable to an under-coupled one, specifically suggesting an optimum of κ opt ext ≈ κint + 4g 2ens/Γ for maximizing sensitivity.

4. Frequency-dependent magnetic anomaly detection: The system can operate as a frequency-sensitive magnetometer, showing that the sensitivity is flat at 93 pT/√Hz up to about 100 kHz and rolls off above the 360 kHz bandwidth set by the spin and mode linewidths.

5. Multi-modal sensing capability: By analyzing three distinct lines exist at any single bias field, the system can resolve three different magnetic field configurations simultaneously, offering a multiplexing gain of 1.72 over a single edge measurement.

Nitrogen-vacancy (NV) centre ensembles are leading room-temperature magnetometers, but their sensitivity is usually extracted from a single resonance and relies on external microwave and bias-field hardware. This work proposes a chip-scale NV magnetometer built around a one-dimensional YIG/CoFeB magnonic crystal, in which a single structural defect forms a passive magnonic cavity and integrated permanent magnets supply the bias field. Micromagnetic simulations show a defect-localised spin-wave mode at 2.910 GHz inside the 2.5-3.0 GHz magnonic band gap, 40 MHz from the NV zero-field splitting, with a band structure that matches an independent plane-wave calculation. A Tavis-Cummings/Lindblad model in which the microwave reaches the NV ensemble through this mode predicts a hyperfine-resolved optically detected magnetic resonance spectrum that is strong only within the mode linewidth, and a dispersive homodyne response with one steep edge per 14 N hyperfine line. The master-equation solutions agree with the closed-form homodyne steady state (R squared > 0.9999) and with an approximate analytic ODMR theory (peak positions within 0.012 MHz). For a critically coupled readout with room-temperature electronics, combining the three edges gives a projected sensitivity of 93 pT/ sqrt Hz, a sqrt 3-fold gain over the best single edge. Mapping the sensitivity against the NV-magnon coupling reveals an optimum near g ens/2π about 0.2 - 0.4 MHz, where it reaches 72 pT/ sqrt Hz for a 0.64 MHz spin linewidth and 19 pT/ sqrt Hz for 0.17 MHz, with thermal limits of 6.6 and 1.7 pT/ sqrt Hz. The antenna-mode coupling, the NV-magnon coupling and the spin linewidth are therefore the main design levers.