A Miniaturized Microwave Magnetometer with High Frequency Resolution Based on Diamond NV Centers for Multi-Microwave-Field Measurement
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Micromachines |
| Published | 2026-05-25 |
| Authors | Yujia Tian, B H Wang, Qiang Zhu, Xin Li, Wenyuan Hao |
| DOI | 10.3390/mi17060647 |
A fiber-coupled diamond nitrogen-vacancy magnetometer measured three microwave magnetic fields simultaneously in a 13 cm³ unit. Its demonstrated frequency resolution was 0.01 Hz, on the order of millihertz, and its sensitivity was 0.385 nT/√Hz.
Continuous heterodyne measurement
Section titled “Continuous heterodyne measurement”NV centers were optically pumped with a 532 nm laser, and their 637 nm fluorescence provided the readout. The ground state had a 2.87 GHz zero-field splitting between the ms = ±1 sublevels. An applied microwave field opened a transverse relaxation channel and reduced fluorescence from the ms = 0 state.
When a reference field E1 and one or more test fields acted together, interference produced sum and difference terms in the fluorescence. The photodetector and lock-in amplifier had bandwidth below 100 MHz, so difference terms, typically below 200 Hz, were retained. The sum terms were filtered out.
Fast Fourier transform analysis separated the retained beat notes. Each test field contributed a beat at its frequency difference from the reference. Beat amplitudes were proportional to the products of the interacting field amplitudes. Random phases were acceptable when each frequency remained stable during acquisition. Rapid phase fluctuations could broaden a spectral line beyond the FFT resolution, and the authors placed that case outside the intended scope.
Diamond and optical assembly
Section titled “Diamond and optical assembly”The sensing element was a high-temperature, high-pressure diamond plate measuring 3 × 3 × 0.2 mm³. It was cut along [100]. The authors stated that this cut improved lattice stability and avoided isotope-related effects on hyperfine splitting. The diamond contained 50 ppm nitrogen. Electron irradiation lasted 2.5 h at a dose of 9.8 × 10^15 cm−2 and 10.0 MeV. A 2 h anneal at 850°C followed.
An 80 mW 532 nm beam entered through an optical fiber and was collimated inside the unit. A dichroic mirror directed it through a circular copper patch antenna before it reached the diamond. The antenna used 0.5 mm-diameter copper wire, had a 5 mm diameter, and sat about 1 mm above the diamond.
Red fluorescence passed back through the dichroic mirror, an optical filter, and a photodiode. A Zurich Instruments HF2LI lock-in amplifier converted the photodiode signal to digital data. FFT processing in PyCharm 2021 extracted the frequency components. The laser, antenna, and collection optics shared the 13 cm³ unit. The microwave source, lock-in amplifier, and computer remained external.
Frequency resolution
Section titled “Frequency resolution”The reference E1 was fixed at 2.86 GHz and 30 dBm while the frequency of E2 was varied. The minimum distinguishable difference was 0.01 Hz, set by the microwave source’s minimum frequency resolution.
The Fourier relation Δf = 1/T sets the acquisition-time limit. The demonstrated 0.01 Hz resolution used a 100 s measurement. A larger beat frequency reduced the fluorescence amplitude because the fluorescence response had a limited rate. The retained difference signals were typically below 200 Hz.
The experiment therefore combined two requirements. Each test field was measured relative to the reference, and the resulting beat had to remain within the low-frequency range retained by the readout electronics. The FFT then separated that beat from the remaining fluorescence noise.
Power response and sensitivity
Section titled “Power response and sensitivity”With the E1 and E2 difference fixed at 100 Hz, the authors reduced E2 power and recorded the fluorescence. Figure 4 covers powers from 1 W to 0.000001 W. The fluorescence amplitude decreased as E2 weakened.
At low power, components above 1.5 kHz appeared. The authors attributed them to residual reference effects, lock-in amplifier harmonics, photodiode noise, and acquisition electronics. They do not affect extraction of the low-frequency beat.
The Rabi frequency was linear with the square root of microwave power. The authors derived a microwave-field coefficient of k = 0.818 from this relation and the fluorescence response. The amplitude spectral density was measured over 100 s. Voltage noise was averaged from 1 to 100 Hz, excluding the 50 Hz line harmonic. This gave a sensitivity of 0.385 nT/√Hz.
The reference field was indispensable because a test field alone produced no oscillatory beat. A 30 dBm reference did not mask test signals down to -20 dBm, according to the authors. The high reference power behaved mainly as an offset while the analysis retained the low-frequency difference component.
Simultaneous three-field detection
Section titled “Simultaneous three-field detection”Adding E3 allowed simultaneous detection of two test fields against E1. Increasing the test frequencies reduced the fluorescence amplitude and made the trace less ordered. The authors attributed this behavior to additional high-frequency microwave components and noise.
In a separate power sweep, the time trace disappeared when E2 and E3 each reached 0.004 W. FFT still detected the weaker signals.
With E1 plus E2 and E3 at 50 Hz and 80 Hz beat frequencies, frequency accuracy was better than 0.01 Hz. Amplitude error was below 2%, crosstalk was below -35 dB, and the SNR = 3 detection limit was around 1.20 nT.
These measurements covered simultaneous operation with different test frequencies and power levels. The frequency-domain result remained useful when the time-domain oscillation was buried in noise.
Comparison with prior magnetometers
Section titled “Comparison with prior magnetometers”The paper compares the prototype with three NV-center magnetometer approaches.
| Approach | Sensitivity (nT/√Hz) | Volume in paper | Function in paper |
|---|---|---|---|
| Fu et al. [21] | 0.93 | Portable | Real-time vector magnetic measurement |
| Wang et al. [20] | 3.58 | ~12 cm³ | High dynamic range magnetic field measurement |
| Mao et al. [19] | 0.34 | 3.81 cm³ | High-sensitivity magnetic field detection |
| This work | 0.385 | 13 cm³ | Multi-frequency microwave magnetic field detection |
Mao et al. had the smallest listed numerical volume at 3.81 cm³. Wang et al. was approximately 12 cm³, while this magnetometer was 13 cm³. The paper describes its 0.385 nT/√Hz sensitivity as competitive while assigning the other systems different functions. Its comparison table does not list frequency-resolution values for the prior devices.
Limits and engineering implications
Section titled “Limits and engineering implications”Frequency separation has a direct time cost. A shorter acquisition gives coarser frequency resolution through Δf = 1/T, while the demonstrated 0.01 Hz resolution required 100 s. The method also requires stable frequencies during that interval.
The reference field cannot be removed because the test field alone causes a static fluorescence reduction rather than an extractable beat. Strong test frequencies further reduced fluorescence amplitude and increased trace disorder. At weak powers, system noise peaks became visible, although FFT still recovered the low-frequency difference component.
The compact unit does not make the complete system self-contained. It still depends on external microwave sources, a lock-in amplifier, and a host computer. The authors plan to use micro or nano fabrication to reduce the sensor size. They also plan to redesign the external equipment and add a microwave focusing structure. The focusing structure was proposed to increase the local microwave field and reduce environmental noise.
The underlying data are not public, but the authors state that they may be obtained upon reasonable request. Material selection must account for the [100] cut, 50 ppm nitrogen concentration, irradiation, and annealing used in the experiment. Device operation must also provide a stable reference field and accept the acquisition time needed for fine frequency separation.
Materials and support
Section titled “Materials and support”For an NV microwave magnetometer, 6CCVD’s diamond materials catalog lists MPCVD-grown single-crystal plates and wafers up to 125 mm. Its stated thickness range of 0.1 to 500 micrometers includes the paper’s 0.2 mm element, but the experiment used HPHT rather than MPCVD diamond. The paper’s [100] cut, 50 ppm nitrogen content, electron irradiation, and annealing require review through engineering support before substituting a catalog plate.
View Original Abstract
Diamond nitrogen-vacancy (NV) centers are regarded as promising microwave sensors owing to their excellent magnetic sensitivity, stability, and environmental compatibility. However, traditional confocal test platforms based on diamond NV centers are bulky, which limits their practical applications. In this paper, a fiber-coupled compact NV microwave magnetometer is designed that employs the continuous heterodyne measurement method and a fast Fourier transform to measure multiple microwave fields. We integrated the laser excitation module, microwave antenna module, and fluorescence collection module into a single unit, reducing the volume of the magnetometer to 13 cubic centimeters. By adjusting the frequency and power of the measured microwave signals, the applicability of the device under different frequency and power conditions was verified. Experimental tests show that the microwave magnetometer can simultaneously detect multiple microwave fields with different frequencies and power levels, achieving a frequency resolution on the order of millihertz (mHz) and a microwave detection sensitivity of 0.385 nT/Hz1/2. These results demonstrate the magnetometer’s multi-microwave-field measurement capability, making it highly promising for applications such as microwave anomaly localization and medical diagnosis.
Single Crystal Cvd Diamond Plate
View Specs →
Boron Doped Single Crystal Diamond Plate
View Specs →
Customized Single Crystal Cvd Diamond Plate
View Specs →References we cover
Section titled “References we cover”- High-Dynamic-Range Integrated NV Magnetometers
- Picotesla magnetometry of microwave fields with diamond sensors
- System for the remote control and imaging of MW fields for spin manipulation in NV centers in diamond
- Noninvasive Imaging Method of Microwave Near Field Based on Solid-State Quantum Sensing
Original source
Section titled “Original source”References
Section titled “References”- 1988 - Optically detected spin coherence of the diamond N-V centre in its triplet ground state [Crossref]
- 1997 - Scanning Confocal Optical Microscopy and Magnetic Resonance on Single Defect Centers [Crossref]
- 2018 - A new B-dot probe circuit for magnetic diagnostics of radio frequency discharges [Crossref]
- 2025 - Microwave photonics promotes emerging integrated sensing and communication technology [Crossref]
- 2018 - Noninvasive Imaging Method of Microwave near Field Based on Solid-State Quantum Sensing [Crossref]
- 2020 - A substitution method for nanoscale capacitance calibration using scanning microwave microscopy [Crossref]
- 2024 - Coaxial tips for a scanning microwave microscope and its calibration with dielectric references [Crossref]
- 2023 - Imaging microwave field of chip surfaces based on scanning microwave microscopy [Crossref]
- 2021 - Real-Time Removal of Topographic Artifacts in Scanning Microwave Microscopy [Crossref]
- 2024 - Dual-cameras terahertz imaging with multi-kilohertz frame rates and high sensitivity via Rydberg-atom vapor [Crossref]