Achieving strain-limited ensemble dephasing times in solid-state quantum sensor with a compact single magnet assembly
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Applied Physics Letters |
| Published | 2026-09-28 |
| Authors | Y. MASUYAMA, Chikara Shinei, Haruki Tanaka, Hiroshi Abe, Takashi Taniguchi |
| DOI | 10.1063/5.0317515 |
A 3 cm cubic passive magnet assembly produced an ensemble dephasing time of 5.32 µs across a 370 µm-thick diamond. The measured value matched a 5.33 µs Helmholtz-coil result while the assembly occupied 0.23% of the coil volume. The remaining dephasing was attributed to strain in the diamond, with a rate of 188 kHz.
Uniform bias field sets the sensing requirement
Section titled “Uniform bias field sets the sensing requirement”An inhomogeneous bias field shifts qubit energies and shortens the ensemble dephasing time. For a whole mm-scale diamond with about 1 ppm substitutional nitrogen, field homogeneity must be within 99.9% to remain P1-limited. The authors relate this requirement to a 2.5 µT change within a 2.5 mT bias field.
Helmholtz coils, magnet pairs, and Halbach arrays can provide uniform fields, but those sources occupy too much space for a portable probe. The authors chose a 5 cm cube as the practical target for the probe. Their assembly instead produced a field difference below 1.8 µT across approximately 1 mm around the sensing region.
Inverse design shaped the magnet array
Section titled “Inverse design shaped the magnet array”The authors used inverse problem analysis to optimize magnet placement for three-dimensional field uniformity. The procedure iteratively calculated a magnetic moment arrangement and evaluated the resulting field distribution. This differs from forward design, which selects a familiar shape and then assesses its field.
The assembly contains 2 mm cubic neodymium magnets arranged in three layers. A 3D-printed support held the magnets in their optimized positions, and a 3D-printed case joined the layers into one assembly. Finite-element calculations predicted field uniformity of at least 99% at several millimeters from the source, comparable to a Helmholtz coil.
Across the modeled 370 µm sample thickness, the calculated field changed by less than 0.4 µT. This variation corresponded to less than 10 kHz in Zeeman shift. The transverse components were three to four orders below the longitudinal field. A perturbation model predicted a resonance shift of several Hz, or tens of parts per billion. The authors found the effect on dephasing negligible.
A Helmholtz coil suppresses transverse components by four to five orders. One 2 mm cubic magnet produced transverse components equal to one-tenth of its longitudinal field. The authors found that this configuration affected dephasing. The multi-magnet arrangement therefore supplied the needed directional control.
Diamond preparation set the material baseline
Section titled “Diamond preparation set the material baseline”NV centers in diamond served as the qubits. The specimen was a (111)-oriented HPHT crystal with 0.79 ppm initial nitrogen and a 99.998% carbon-12 ratio. Electron irradiation used 2 MeV energy and a dose of 5.0 × 10^16 cm⁻². Annealing at 1375 ± 25 °C in vacuum created the NV centers. Electron spin resonance estimated their concentration at 0.05 ppm.
The assembly applied approximately 2.5 mT along the (111) axis. This field selectively addressed the $m_s = -1$ component. The authors note that frequency-shift variations remain negligible when detection depth is limited to approximately 1 µm, as in confocal microscopy. The 370 µm sample instead produced non-negligible shifts across its full thickness.
Columnar excitation isolated magnet-induced dephasing
Section titled “Columnar excitation isolated magnet-induced dephasing”Columnar excitation fluorescence microscopy evaluated the assembled sensor under realistic sample volume and microwave power density. Its long focal depth was approximately 20 µm at the focal position, which provided uniform excitation through the diamond thickness.
At each magnet position, the authors measured the NV resonance with pulsed ODMR. They calibrated the π-pulse duration through Rabi oscillations and then measured $T_2^*$ with a Ramsey sequence. A microwave field detuned by −3.2 MHz resolved the oscillatory components associated with the nitrogen-14 hyperfine structure.
The fit used three oscillatory components and an offset. A representative waveform gave $T_2^*$ of 5.21 ± 0.12 µs, with component frequencies of 0.85, 3.0, and 5.2 MHz. The motorized stage moved only the magnet, leaving optical excitation and microwave irradiation fixed. This arrangement isolated dephasing caused by field inhomogeneity.
The maximum $T_2^$ occurred at a 6.6 mm gap between the sample and magnet surface. Outside this optimized region, field uniformity declined and $T_2^$ fell.
Helmholtz coil comparison confirmed the field contribution
Section titled “Helmholtz coil comparison confirmed the field contribution”| Measurement | Compact assembly | Helmholtz coil |
|---|---|---|
| Average $T_2^*$ | 5.32 ± 0.22 µs | 5.33 ± 0.29 µs |
| Sampling | 100 sequential measurements | Continuous Ramsey measurements |
| Position | 6.6 mm gap | Same sample position |
| Geometry | 3 cm cube assembly | 300 mm outer diameter, 230 mm inner diameter |
| Power | Passive | Powered coil |
The Helmholtz coil had field uniformity above 99.92% over a 5 mm cube. Strain set its 188 kHz dephasing rate, and the same strain limit applied to the compact assembly. The authors describe this rate as the state of the art for bulk quantum sensors.
The compact assembly added less than 50 kHz of magnetic-field dephasing over a 0.6 mm range. The authors state that this condition held across approximately 1 mm, including the 370 µm sample. A 50 kHz rate corresponds to a 1.8 µT difference between maximum and minimum Zeeman shifts. That equals 720 ppm of 2.5 mT and matches the ideal P1-limited value for 0.5 ppm substitutional nitrogen.
Repeated compact magnet measurements showed no significant time-dependent drift. The coil data had larger deviations, which the authors attribute likely to power-supply noise. The passive assembly removes that noise source.
Alignment tolerance and model limits
Section titled “Alignment tolerance and model limits”XY scans measured $T_2^$ at gaps of 6, 6.6, and 7 mm while the magnet moved across the plane parallel to its surface. The region with $T_2^$ above 4.5 µs covered roughly 1 × 1 mm². The authors state that the current 0.6 mm positioning tolerance lies within standard machining and optical assembly precision.
The analysis combined COMSOL Multiphysics finite-element fields with a one-dimensional Ramsey model. Setting the sample thickness to 367 µm reproduced the measured distance dependence. The model also calculated the theoretical upper limit from magnet-induced inhomogeneity alone. Its $T_2^*$ matched the value associated with a P1 concentration of 17 ppb. The authors explicitly present this as a theoretical limit rather than an experimental demonstration.
Device implications and remaining limits
Section titled “Device implications and remaining limits”For device engineering, material selection and magnet placement must be designed together. The paper identifies spin impurities, carbon-13, strain, temperature inhomogeneity, and applied-field homogeneity as limits on ensemble dephasing. Its result shows why $T_2^*$ should be measured across the intended sensing volume rather than at one focal depth.
Two practical constraints shape the module. Placement must remain within the accepted 0.6 mm tolerance because transverse displacement shifts the $T_2^*$ distribution. Standard NdFeB magnets also drift with temperature and shift qubit energies. Diamond’s high thermal conductivity makes the temperature shift uniform, so drive-frequency adjustment can compensate it. A Ramsey measurement takes about 1 min, which limits temperature drift during signal integration. The authors suggest SmCo as an alternative for harsher environments.
External magnetic noise still requires a system-level approach such as a gradiometer. The passive assembly removes coil supply noise but does not reject external fields. Scaling the assembly relative to the diamond could extend the uniform sensing region beyond the present 1 mm dimensions. That extension remains a suggested design direction, not a measured device result.
Materials and support
Section titled “Materials and support”A compact NV magnetometer needs a single-crystal plate, but the paper used a 370 µm-thick (111)-oriented HPHT crystal with 99.998% carbon-12 and 0.05 ppm NV centers. The 6CCVD diamond materials catalog offers MPCVD single-crystal plates from 0.1 to 500 µm thick, which includes the paper’s thickness. An engineering support review should check the growth route, (111) orientation, carbon-12 enrichment, and NV concentration before substituting the catalog material.
View Original Abstract
Solid-state quantum sensors offer the promising prospect of high-sensitivity sensing across various environments, including at room temperature and atmospheric pressure. Highly sensitive solid-state quantum sensors require long ensemble dephasing times (T2*) in mm-scale sample areas. This requirement demands a highly uniform bias magnetic field across all qubits, which has traditionally required large magnetic field sources such as Helmholtz coils or a pair of magnets. Meanwhile, to expand the range of applications of quantum sensors, portable sensors are required, but it is difficult to carry these large magnets. Balancing the miniaturization of the quantum sensor with high sensitivity is a significant challenge. In this study, we achieve a long ensemble dephasing time for nitrogen-vacancy (NV) centers in a large sample volume using a single, compact, and optimized magnet assembly. By using a magnet design derived from an inverse problem-based analysis, we reduced the magnet assembly volume to 0.23% of the Helmholtz coil volume. We achieve an ensemble dephasing time of 5.32 μs for the entire NV center ensemble across the sample thickness, comparable to that obtained with a standard Helmholtz coil setup. This dephasing is limited by strain-induced dephasing across the diamond sample thickness, with a dephasing rate of 188 kHz, which represents the state-of-the-art diamond quality for bulk quantum sensors. Our work paves the way for the miniaturization of high-sensitivity quantum sensors, enabling new sensor designs and the construction of compact, highly sensitive quantum sensor arrays.
High Concentration Quantum Grade Single Crystal Diamond
View Specs →
Single Crystal Cvd Diamond Plate
View Specs →
Boron Doped Single Crystal Diamond Plate
View Specs →References we cover
Section titled “References we cover”- Miniaturized magnetic-field sensor based on nitrogen-vacancy centers
- Homogeneous spin-dephasing time of NV− centre in millimetre-scale 12C-enriched high-pressure high-temperature diamond crystals
- Compact and Fully Integrated LED Quantum Sensor Based on NV Centers in Diamond
- Nitrogen related paramagnetic defects - Decoherence source of ensemble of NV− center
- Ensemble Negatively-Charged Nitrogen-Vacancy Centers in Type-Ib Diamond Created by High Fluence Electron Beam Irradiation