Skip to content

Real-Time Magnetic Field Sensing based on Microwave Frequency Modulated Photocurrent of Nitrogen-Vacancy Centers in Diamond

FieldValue
JournalOpen MIND
Published2026-02-08
AuthorsXuan-Ming Shen, Qilong Wu, Huihui Yu, Pei-Nan Ni, Qing Lou
DOI10.48550/arxiv.2602.07926

Microwave frequency-modulated photocurrent detection tracked a 1 Hz alternating magnetic field with a standard deviation of 1.5 μT. It also measured a magnetic sensitivity of 921 nT/√Hz from 1 to 10 Hz.

Frequency-modulated PDMR provided the real-time readout and the measured 921 nT/√Hz sensitivity. All frequency-modulation measurements used 50 V DC bias, 100 mW laser power, 26 dBm microwave power, and 1 MHz modulation depth.

Readout modeMeasurement conditionsResult
Laser intensity modulation50 V bias, 100 mW laser, 26 dBm microwave, 321 Hz square wave397 nT/√Hz theoretical shot-noise limit
Microwave frequency modulation50 V bias, 100 mW laser, 26 dBm microwave, 1 MHz modulation depth898 nT/√Hz theoretical and 921 nT/√Hz measured from 1 to 10 Hz
Real-time tracking50 V bias, 100 mW laser, 26 dBm microwave, electromagnet varied at 1 Hz0 to 27 μT with 1.5 μT standard deviation

The measured value was about twice the 397 nT/√Hz estimate from the laser-modulated configuration. Earlier PDMR magnetometry studies mainly estimated sensitivity through formulas rather than validating it with amplitude spectral density. To their knowledge, the authors identify the 1 Hz tracking as the first real-time PDMR magnetic-field demonstration.

Photoelectric detection of magnetic resonance, or PDMR, reads the NV spin state through light-induced charge conversion. Electrical readout removes the need to collect NV fluorescence and convert it into an electrical signal.

Under 532 nm illumination, conversion between NV⁻ and NV⁰ releases electrons and holes. This produces a spin-dependent photocurrent in the picoampere range. A microwave near 2.87 GHz transfers population from the (m_s=0) state to the (m_s=\pm1) states and reduces the photocurrent. A current preamplifier and lock-in amplifier resolve this reduction.

Laser intensity modulation produced a dip-shaped PDMR spectrum. Microwave frequency modulation produced a dispersive spectrum with a linear region near 2870 MHz. The lock-in voltage in that region changed linearly with microwave carrier frequency. This response allowed direct conversion from lock-in voltage to magnetic field.

Device fabrication and electrical response

Section titled “Device fabrication and electrical response”

The sample was MPCVD-grown diamond with two pairs of gold interdigitated electrodes and a gold microbelt microwave antenna. Each interdigitated electrode was 10 μm wide with 10 μm spacing. The antenna strip was 25 μm wide and 15 μm from the electrodes.

After acid and ultrasonic cleaning, a 100 nm gold film was deposited by magnetron sputtering. Laser direct writing and photolithography formed the electrode and antenna patterns. The finished device was annealed at 510 °C under vacuum for 3 hours to lower the diamond-electrode interface barrier.

The authors modeled the electrode-diamond-electrode structure as two back-to-back Schottky junctions. At high bias, their depletion regions merged into one effective junction. The low-voltage branch saturated at 1.5 V, and the high-voltage branch had a fitted threshold of 18.28 V. The two fitted current amplitudes were 59.1 ± 2.3 pA and 93.31 ± 7.1 pA. Measurements used 50 V because the high-bias photocurrent was easier to detect with the lock-in amplifier.

At 50 V, photocurrent contained a linear term of 6.602 pA/mW and a quadratic term of 0.079 pA/mW². The terms were equal near 86 mW, so the authors used 100 mW for the NV response.

The 100 mW laser was square-wave modulated at 321 Hz. Photocurrent passed through a 1 nA/V transimpedance amplifier and a 0.3 to 1 kHz band-pass filter. The lock-in integrated near-resonant and detuned photocurrent for 4 s to calculate contrast.

A Lorentzian fit gave 12.1% maximum contrast, a 2869 MHz center, and a 19.5 MHz linewidth. The fitted signal and carrier rate gave a theoretical shot-noise-limited sensitivity of 397 nT/√Hz. The dip spectrum did not scale linearly with microwave frequency, limiting direct conversion of the lock-in output to field.

Microwave frequency modulation and real-time tracking

Section titled “Microwave frequency modulation and real-time tracking”

Frequency modulation changed the measured PDMR line from a dip to a dispersive trace. Under the main operating settings, the fit gave a 2868.43 MHz center and a 15.257 MHz linewidth. The fitted slope was 2.018 × 10⁻⁴ V/MHz. The authors used an electron gyromagnetic ratio of 2.8 MHz/Gauss to convert voltage to field.

The theoretical sensitivity was 898 nT/√Hz with a 10 Hz equivalent noise bandwidth. To test it, they recorded lock-in voltage for 5 s at resonance and converted the trace to magnetic field. One-second segments were analyzed by discrete Fourier transform to calculate amplitude spectral density.

The measured sensitivity was 921 nT/√Hz from 1 to 10 Hz. This was close to the 898 nT/√Hz estimate. It was about twice the 397 nT/√Hz value from the other estimation formula. The authors attribute that difference to electronic noise in the measurement setup.

An electromagnet beside the setup supplied a 1 Hz alternating field. The lock-in voltage alternated between 0.15 mV and 0 mV, corresponding to 27 μT and 0 μT. The final 1 s segment had a mean of 27 μT and a standard deviation of about 1.5 μT.

At 26 dBm microwave power, raising laser power from 70 to 130 mW changed the contrast. Contrast rose from 11.1% to 12.1% at 100 mW, then fell to 10.3%. The linewidth increased from about 14.1 MHz to 21.9 MHz. Estimated sensitivity fell from 600 nT/√Hz to about 400 nT/√Hz by 100 mW, then saturated.

At 100 mW laser power, raising microwave power from 20 to 26 dBm increased contrast from 6% to 12%. The linewidth increased from 14 MHz to 20 MHz, while estimated sensitivity decreased from 550 to 400 nT/√Hz. The authors linked these changes to stronger stimulated microwave absorption. At high laser power, contrast reflected competition between optically induced spin polarization and microwave absorption.

The master equation included charge-state dynamics, optical pumping, spin transitions, and about 2,551 illuminated NV centers. It also linked laser power to the optical pumping rate and microwave power to the local microwave field. Because field distribution and NV position relative to the antenna were difficult to estimate, the local microwave field was treated as a fitting parameter. The simulated contrast, linewidth, and sensitivity followed the measured power dependence.

Confocal fluorescence mapping estimated an NV concentration of 3.3 ppb. Spin characterization and a literature relation gave 1.5 ppm nitrogen and an NV⁻/N ratio of 2.2 × 10⁻³.

The authors propose electron irradiation to raise NV concentration by three orders of magnitude, reaching a few ppm. The model predicts a 32-fold sensitivity improvement to tens of nT/√Hz if carrier detection rises with NV concentration. This projection was not tested with an irradiated sample.

The demonstrated setup used 50 V bias, 100 mW optical power, and 26 dBm microwave power. The 1.5 μT standard deviation applies to a tracked 27 μT field at 1 Hz. It does not equal the 921 nT/√Hz noise sensitivity. Electronic noise therefore remains a visible limit on the frequency-modulated sensor.

For device engineering, the paper identifies NV concentration, local microwave coupling, and carrier generation as coupled design variables. The model supports higher-concentration diamond as a route to lower noise, but experimental noise validation remains necessary.

The paper’s device requires MPCVD-grown diamond with Au electrodes and a patterned microwave antenna for NV-center photocurrent sensing. 6CCVD’s diamond materials catalog lists MPCVD-grown single-crystal and polycrystalline plates and wafers up to 125 mm, with Au among the metallization options. The engineering support page can define the custom fabrication requirements for the electrode and antenna patterns.

View Original Abstract

While photoelectric detection of magnetic resonance (PDMR) can be applied to miniaturize nitrogen-vacancy (NV) center-based quantum sensors, real demonstration of PDMR-based magnetic field sensing remains as a distinctive challenge. To tackle this challenge, in this article, we fabricate diamond samples with electrodes and microwave antenna on the surface, and realize PDMR by detecting photocurrent in picoampere range via various lock-in amplifying modes. We obtain a theoretical and experimental sensitivity 397 nT/Hz and 921 nT/Hz of magnetic field detection in DC-10 Hz range with a laser intensity and microwave frequency modulation mode, respectively, and demonstrate for the first time, a real-time tracking of alternating magnetic field with a standard deviation of 1.5 uT. Furthermore, we investigate systematically the dependence of the PDMR contrast, linewidth and the sensitivity on the laser and microwave power, and find a perfect agreement with a master equation-based theory. Thus, our results represent a critical step forward in transitioning PDMR from a spectroscopic technique to a practical sensing modality.