Skip to content

Integrated fiberized diamond magnetometer with high sensitivity and dynamic range by magnetic flux concentrators

FieldValue
JournalCommunications Physics
Published2026-07-27
AuthorsXuedong Gao, Xicheng Zhang, Cui Yu, Yu Cui, Mao Ze-Kun
DOI10.1038/s42005-026-02779-1

A fiber-coupled nitrogen-vacancy diamond magnetometer reached 390(172) fT/√Hz above 3 Hz while tracking fields to 180 µT, about four times Earth’s magnetic field. Shielded closed-loop measurements also gave 50 pT/√Hz at 0.01 Hz. The authors combined a [111]-oriented nitrogen-doped diamond element, two magnetic flux concentrators, transverse pump polarization, and closed-loop tracking of two optically detected magnetic resonance transitions.

Sensor construction and material preparation

Section titled “Sensor construction and material preparation”

The sensing element was diamond grown by microwave plasma chemical vapor deposition with a [111] crystal orientation. Its initial nitrogen concentration was about 48 ppm. Electron irradiation used 10 MeV energy and a dose of 5 × 10^18 e−/cm². A 4 h anneal at high temperature and high vacuum followed. The resulting NV concentration was about 8 ppm.

Laser cutting produced elements about 220 × 220 × 110 µm³. Their upper faces were polished perpendicular to the [111] direction. Under about 319 mW of 532 nm pumping, the integrated probe collected approximately 1.45 mW of red fluorescence. The excitation beam traveled through a polarization-maintaining fiber with a 3.5 µm core and 125 µm cladding. It reached the diamond from the side, allowing the pump polarization to remain perpendicular to the measured NV axis.

The diamond sat between two magnetic flux concentrators. Their amplified field was parallel to the [111] NV axis. The measured ODMR resonance shift gave a magnetic field magnification of about 360. Simulation gave gains between 320 and 450. Gain nonuniformity broadened the ODMR spectrum. At 2 µT, the simulated broadening was about 7.28 MHz. The excitation area was smaller than the diamond, so actual broadening at higher fields was smaller than the simulation.

Perpendicular pump polarization gave the highest excitation efficiency for the [111] center. It also suppressed emission from other axial centers and increased ODMR contrast. The [111] axis had to align with the amplified field within 0.1°. At the geomagnetic field, the resulting transverse component was about 0.3 G. The authors state that ODMR broadening could then be ignored.

The concentrator gain changed by only about 0.0019% for an 80 K temperature increase. The zero-field splitting was more temperature-sensitive. A 1 K change corresponded to about 2678 nT. The authors used dual-frequency measurements to suppress this drift and calibrate temperature effects.

Resonance readout and closed-loop tracking

Section titled “Resonance readout and closed-loop tracking”

FM-ODMR converted magnetic field changes into resonance frequency shifts. The maximum zero-crossing slope reached 14.7 V/GHz at a 4 MHz frequency deviation. The slope fell at larger deviations. The authors suggest that the excess deviation exceeded the ODMR linewidth.

A lock-in amplifier demodulated the fluorescence. A PID controller then switched a frequency-agile microwave generator to keep FM-ODMR at its zero crossing. This closed loop tracked two NV transitions in real time. Sensitivity was calculated from amplitude spectral density using the ODMR slope, the 28 GHz/T NV gyromagnetic ratio, the MFC gain, and a 4/3 dual-resonance factor.

The closed-loop system reached 50 pT/√Hz at 0.01 Hz and 390(172) fT/√Hz above 3 Hz. Both values were measured in a magnetic shielding environment. The open-loop results and calculated noise floors provide context.

Operating conditionFrequency conditionSensitivity or calculated limit
Closed-loop magnetometer0.01 Hz50 pT/√Hz
Closed-loop magnetometerAbove 3 Hz390(172) fT/√Hz
Open-loop resonance readout0.01 Hz7.5 nT/√Hz
Open-loop resonance readoutAbove 3 Hz2.8(1.5) pT/√Hz
Shot-noise calculation with MFCsHigh-sensitivity regimeApproximately 139 fT/√Hz
Modeled permalloy magnetization noiseAbove 3 HzBelow 205 fT/√Hz

Open-loop sensitivity worsened as the measured field increased. The narrow-bandwidth microwave antenna reduced the FM-ODMR zero-crossing slope at higher fields. The closed-loop system extended the detection range to 180 µT. ODMR contrast decreased slowly across that range, which the authors compare favorably with earlier [100] and [110] diamond measurements.

Closed-loop tracking reduced the influence of laser power fluctuations. The remaining low-frequency noise limited the 0.01 Hz result. The paper attributes this residual 1/f noise to the concentrator’s inherent noise and thermal effects. Imperfect compensation of the zero-field splitting and temperature-dependent changes in the concentrator gap also contributed.

The cited modular magnetometer reached 0.3 ± 0.2 nT/√Hz. Fiber-coupled probes without MFCs had reported sensitivities spanning 12 pT/√Hz to 4.2 nT/√Hz. The best cited MEMS sensitivity was 334 pT/√Hz. The smallest cited CMOS magnetometer had a volume of 1.5 mm³.

Two earlier concentrator studies used either MN60 ferrite or permalloy. Fescenko and co-workers reported 0.9 pT/√Hz with two ferrite concentrators. Xie and co-workers reported 196 fT/√Hz with two permalloy concentrators under a non-resonant condition. Their resonant measurement gave approximately 500 pT/√Hz.

The paper’s room-temperature comparison places this magnetometer at the high-sensitivity end of the reported systems. The authors position it as retaining sub-pT high-frequency sensitivity while measuring to a range of about four times Earth’s field. Earlier concentrator approaches were described as suffering from ODMR broadening and restricted dynamic range.

The portable package combined the probe, a fiber-coupled 532 nm laser, frequency-tracking modules, and a portable battery. Without magnetic shielding, the closed-loop system tracked an AC magnetic field with an amplitude of about 15 µT at 1 Hz. The two transitions were sampled at approximately 200 Hz. This test confirmed large-range tracking in an unshielded environment.

The magnetometer also monitored a high-speed railway power network from approximately 120 m away. The test track section was about 50 km long. A train entered the power supply section four seconds after monitoring began. The locomotive load produced a detectable 50 Hz AC magnetic field.

Three passing trains caused low-frequency field deviations of approximately 5 nT after low-pass filtering. The AC amplitude increased as a train approached. A transient decrease occurred at 62 to 64 s when the train center passed directly above the sensor. The authors attribute this decrease to cancellation between opposing fields from the catenary wire and rail return path.

The concentrator volume constrains spatial resolution. When the target is much smaller than the concentrator, the field develops a substantial gradient. The amplification factor at the diamond then decreases, and the field lines become distorted. The authors therefore identify biomagnetic measurements as a difficult application for the present probe geometry.

The theoretical range is limited by magnetic saturation in the MFCs and the microwave source’s output frequency range. The authors suggest that a MEMS process could reduce probe size. They also propose using a 12C-purified diamond sample to reduce ODMR linewidth. Neither improvement was demonstrated in this work.

The sensor can isolate the field component along a selected axis. The authors state that this rejects interference from orthogonal directions. The concentrator design also avoids the external bias field used in some other NV magnetometers, preventing bias-induced temperature drift and magnetic noise.

The magnetometer requires a nitrogen-doped, [111]-oriented MPCVD single-crystal diamond element with polished faces for NV fluorescence collection. 6CCVD’s diamond materials catalog lists single-crystal diamond grown by MPCVD and polishing to Ra below 1 nm, but those facts do not confirm nitrogen content or [111] orientation. Engineering support can be used to check the exact material specification before design-in.

View Original Abstract

Magnetometers based on nitrogen-vacancy (NV) centers in diamonds have been widely studied. With the use of purified diamond and magnetic flux concentrators (MFCs), the sensitivity of magnetic field measurement has been improved to the subpico-tesla level. However, diamond magnetometers face the challenge of achieving high sensing performance after integration. Here, we demonstrate an integrated fiber-coupled diamond magnetometer based on an MFC with both high sensitivity and a wide dynamic range. We sandwiched a nitrogen-doped diamond film of [111] crystal orientation between the two MFCs. With the polarization direction of the pump laser perpendicular to the NV center axis, the signal-to-noise ratio and its linearity with the magnetic field are enhanced. We demonstrate a sensitivity of 50 $${{\rm{pT}}}/\sqrt{{{\rm{Hz}}}}[email protected] Hz and 390(172) $${{\rm{fT}}}/\sqrt{{{\rm{Hz}}}}$$@ > 3 Hz while with a detection range of about 4 times the Earth’s magnetic field. This work demonstrates the feasibility of portable, highly sensitive diamond magnetometers that maintain a large dynamic range in various natural environments. Diamond magnetometers leveraging nitrogen-vacancy centers offer precise magnetic field measurements but struggle with integration and miniaturization. Here, the authors present an integrated fiber-coupled diamond magnetometer using magnetic flux concentrators, achieving 50 $${{\rm{pT}}}/\sqrt{{{\rm{Hz}}}}$$ sensitivity and a wide dynamic range, paving the way for portable, high-performance sensing in diverse environments.