Skip to content

Diamond-based magnetometer aboard the International Space Station

FieldValue
JournalPhysical Review Applied
Published2026-03-03
AuthorsYarne Beerden, Boo Carmans, Remy Vandebosch, Dries Hendrikx, Sam Bammens
DOI10.1103/483m-8hfc
Citations1

The OSCAR-QUBE magnetometer operated aboard the International Space Station for 10 months and mapped Earth’s magnetic field in three dimensions. The system used nitrogen-vacancy centers in diamond and achieved a measured sensitivity below 300 nT/√Hz. It processed 231 GB of in-orbit data into field maps that agreed closely with the CHAOS-7 geomagnetic model. The authors also report no significant degradation during long-duration operation in low Earth orbit.

The authors describe OSCAR-QUBE as the first fully integrated, portable NV diamond magnetometer tested in space. The paper positions the instrument below state-of-the-art benchtop sensitivity while retaining low size, weight, and power.

ParameterValueCondition or material
Size1U10 × 10 × 10 cm³
Weight420 gFlight instrument
Power5 WPeak consumption
DC sensitivity<300 nT/√HzIntegrated magnetometer
Measurement rate<40 HzVector mode
Measurement rate<1.3 kHzSingle-axis mode
Dynamic range1.86 mTFlight instrument
Nitrogen concentration14 ppm<100> CVD diamond
NV concentration2 ppm<100> CVD diamond

Most flight measurements used 27 to 29 mW laser power and +26 dBm microwave power over 2660 to 3090 MHz. Sensor parameters were adjusted several times, so the data include operation under different settings. These are system-level figures, not measurements from an isolated sensor head.

The paper cites NV magnetometers with linear response up to 0.1 T, while OSCAR-QUBE lists a dynamic range of 1.86 mT.

Green laser light excites negatively charged nitrogen-vacancy centers, which emit red photoluminescence with spin-dependent intensity. Applied microwaves drive spin transitions, and Zeeman splitting of those transitions encodes the magnetic field. Four tetrahedral NV axes supply the projections needed for three-axis reconstruction.

The mission sample was an electron-irradiated CVD diamond with <100> lattice orientation, 14 ppm nitrogen, and 2 ppm NV centers. The authors selected a high-density NV ensemble to increase photoluminescence signal and signal-to-noise ratio.

Contrast, resonance linewidth, photon rate, and microwave-field uniformity determine continuous-wave ODMR sensitivity. The calculated sensitivity varied from 6.90 to 7.60 nT/√Hz, with a monthly-average standard deviation of ±0.26 nT/√Hz. The flight system’s sensitivity was below 300 nT/√Hz, compared with 6.90 to 7.60 nT/√Hz for the calculated sensor sensitivity.

Compact systems cited by the authors reach subnanotesla sensitivity, but most cited examples stop at sensor-head integration. OSCAR-QUBE instead included control, laser, sensing, microwave, and readout functions in one flight instrument.

Four stacked PCBs separate control and power, laser, diamond sensing, and optical readout functions. An arm holds a 5 × 5 × 1.2 mm³ NdFeB magnet near the diamond to induce Zeeman splitting. The laser housing acts as a heatsink and focusing mount, while an antenna above the diamond applies the microwave field.

Vibration and thermal-vacuum testing found no measurable change in optical alignment or calibration. The payload operated inside the ISS ICECubes Facility, where temperature was controlled from 20 to 25 °C. The ISS orbit averaged 400 to 420 km with a 51.6° inclination; downsampled 1 Hz data had 7.7 km spatial resolution.

Microgravity reduced convective effects, but the orbital day and night cycle still caused thermal fluctuations. Those fluctuations could shift the ODMR line and alter the bias magnet’s alignment or magnetization.

Each packet contained ODMR sweeps, reference magnetometer data, and housekeeping data. Peak detection extracted resonance frequencies for all four NV orientations, followed by least-squares inversion into the sensor frame. Reference-frame transformations moved the vector through the ISS, J2000, Earth-fixed, and North-East-down coordinates. Time-stamped attitude and location data were required for correct comparison with geomagnetic models.

The ground pipeline sorted about 99 million ODMR sweeps, then downsampled measurements to 1 Hz by averaging. CHAOS-7 and IGRF values were appended at matching times for comparison. Axis offsets were then fitted to IGRF because no full in-orbit sensor calibration was performed.

Only 32.3% of acquired magnetic-field data remained usable after filtering. About 45% lacked attitude or location information, and about 20% used incorrect laser or microwave settings. Measurements from 01:00 to 02:00 GMT were also rejected during ICF virus scans because fan operation changed temperature and caused vibration.

Reported ODMR contrast stayed between 1.56% and 1.57% across the eight-month series shown in the paper. Its overall standard deviation was ±0.015%, indicating minimal drift. Resonance FWHM ranged from 9.68 to 9.96 MHz across the mission. Its overall standard deviation was ±0.09 MHz.

The spectra showed minimal resonance shifts and no obvious linewidth change over the mission. The authors therefore observed no significant performance degradation during the reported orbital campaign.

Late in the mission, laser power was reduced from 29 to 27 mW to avoid readout saturation. The lower power reduced red fluorescence and altered sensitivity, producing a baseline shift during later measurements. Laser-temperature trends showed a similar spread to daily total-field values, so the authors identify thermal control as an unresolved error source.

The measured total field ranged from 20 to 54 µT and reproduced large-scale features in CHAOS-7. The South Atlantic anomaly was resolved in the processed map. The authors used this agreement to validate the vector measurements.

ComponentMean residual in nTRMSE in nTPCCWithin ±2000 nT
Total701.461248.400.99288.91%
X299.351205.080.99392.87%
Y−1136.181653.910.95674.34%
Z−952.611068.231.097.52%

The total field had a Pearson coefficient of 0.992 and an RMSE of 1248.40 nT against CHAOS-7. About 88.91% of total-field measurements fell within ±2 µT of the model.

The Y component had the weakest agreement, with an RMSE of 1653.91 nT and 74.34% within ±2 µT. The Z component had a Pearson coefficient of 1.0, although its mean residual and RMSE remained substantial.

The CHAOS-7 dataset had a mean of 33.8 µT, while measurements showed histogram peaks near 33.7 and 36.4 µT. That spread tracked laser temperature, which the authors treat as a possible cause rather than a proven correction.

The system’s sensitivity limits absolute agreement in individual vector components. An expected 4.09° pointing error arose from the ground Helmholtz coil calibration. No zero-field ground calibration was performed, so offsets and a 3 × 3 transformation matrix were inferred from in-orbit data. The authors caution that this indirect calibration adds error and that natural geomagnetic drift biases model-based offsets.

The authors list calibration, temperature, spacecraft noise, axis sensitivity, orientation, and local field effects as possible causes of discrepancies. The ISS magnetic cleanliness was not characterized, so its contribution to the residuals cannot be separated. Brief data losses also occurred during EPS rack downtime and communications interruptions.

For future flights, the authors suggest temperature correction, bias coils, improved in-orbit calibration, and sensitivity optimization. PDMR hardware was included, but the mission used continuous-wave ODMR after final sample selection.

For diamond selection, defect concentration, crystal orientation, linewidth, contrast, and photoluminescence stability matter as much as bulk material quality. A supplier should qualify defect-engineered CVD material against the optical, microwave, thermal, and bias-field design.

The material requirement is <100> CVD diamond with 14 ppm nitrogen and 2 ppm NV centers for an ISS three-axis magnetometer. The flight sample was electron irradiated, while 6CCVD’s diamond materials catalog lists single-crystal MPCVD-grown diamond with plates and wafers up to 125 mm. The engineering support link covers custom fabrication support for mission-specific size, defect-density, and irradiation qualification.

View Original Abstract

Precise mapping of the geomagnetic field is essential for understanding Earth’s geodynamics, space weather interactions, and navigation applications. Traditional magnetometers face limitations in sensitivity for a wide dynamic range, and compactness. To address these challenges, the OSCAR-QUBE quantum magnetometer based on nitrogen-vacancy centers in diamond was developed, providing a sensitive and compact solution for space-based magnetic-field measurements. Our system employs optically detected magnetic resonance to measure magnetic fields utilizing the quantum properties of nitrogen-vacancy centers in a miniaturized design. The form factor of the final device was 1U ( <a:math xmlns:a=“http://www.w3.org/1998/Math/MathML” display=“inline”> <a:mn>10</a:mn> <a:mo>×</a:mo> <a:mn>10</a:mn> <a:mo>×</a:mo> <a:mn>10</a:mn> <a:mspace width=“0.1em”/> <a:msup> <a:mi>cm</a:mi> <a:mn>3</a:mn> </a:msup> </a:math> ), weighing 420 g, and had a power consumption of 5 W. Deployed aboard the International Space Station, our magnetometer measured high-resolution magnetic-field maps, achieving a sensitivity of <d:math xmlns:d=“http://www.w3.org/1998/Math/MathML” display=“inline”> <d:mo><</d:mo> <d:mn>300</d:mn> <d:mspace width=“0.1em”/> <d:mi>nT</d:mi> <d:mo>/</d:mo> <d:msqrt> <d:mi>Hz</d:mi> </d:msqrt> </d:math> and successfully demonstrating vector magnetic-field mapping under low-Earth-orbit conditions. These results validate the flight-proven application of diamond quantum sensing in space, demonstrating the feasibility of solid-state quantum magnetometry for next-generation remote sensing and Earth observation missions. This work lays the foundation for future compact, multisensor quantum payloads for both scientific and commercial space applications.