Skip to content

Quantitative Detection of Molecular Oxygen in the Gas Phase with Fluorescent Nanodiamonds

FieldValue
JournalAdvanced Functional Materials
Published2026-07-14
AuthorsNicholas A. Nunn, Antonin Marek, Alex I. Smirnov, Olga A. Shenderova, Marco D. Torelli
DOI10.1002/adfm.76946

Optically detected magnetic resonance contrast from negatively charged nitrogen-vacancy centers in fluorescent nanodiamonds fell linearly as gas-phase O2 increased from 0 to 100% at ambient pressure. The full-range double-modulated readout gave a raw sensitivity coefficient of (-10.1 ± 0.3) × 10^-4 % mmHg^-1. Its projected detection limit was about 8 mmHg O2, or about 1% in the gas mixture. The setup also detected transient oxygen released by catalase decomposition of hydrogen peroxide.

Continuous-wave ODMR spectra were recorded as the O2 fraction increased in 10% steps. This covered 0 to 760 mmHg O2 partial pressure at a total flow of 200 mL/min. The raw contrast slope was (-7.02 ± 0.33) × 10^-4 % mmHg^-1. The fitted contrast under pure nitrogen was 4.53 ± 0.02%. Normalizing the slope by that intercept produced (-1.55 ± 0.07) × 10^-4 mmHg^-1.

The double-modulated method produced a similar normalized slope of (-1.73 ± 0.05) × 10^-4 mmHg^-1.

ODMR modeO2 range, %Raw slope, % mmHg^-1Normalized slope, mmHg^-1Projected limit, mmHg
CW0 to 100(-7.02 ± 0.33) × 10^-4(-1.55 ± 0.07) × 10^-4N/A
Modulated0 to 100(-10.1 ± 0.3) × 10^-4(-1.73 ± 0.05) × 10^-49
Modulated0 to 21(-14.9 ± 1.0) × 10^-4(-2.53 ± 0.17) × 10^-46
Modulated0 to 5(-22.4 ± 2.6) × 10^-4(-3.81 ± 0.44) × 10^-44
Modulated5 to 1.1(-12.4 ± 0.3) × 10^-4(-2.12 ± 0.06) × 10^-47

The smallest detectable contrast change was 9 × 10^-3%, based on 3σ and a window-average σ of 3 × 10^-3%. This change corresponded to a projected 9 mmHg limit for the full-range sweep. Low-range measurements gave projected limits of 4, 6, and 7 mmHg. The average across all estimates was about 6.5 mmHg. The authors summarized the setup’s limit as approximately 8 mmHg, equivalent to about 1% O2.

The sensor used fluorescent diamond particles approximately 70 nm in diameter with an estimated 2 ppm of negatively charged NV centers. The authors injected 100 µL of a 30 mg/mL aqueous suspension into a glass-bottom channel measuring 0.25 mm by 5 mm by 50 mm. Heating at 70 °C overnight left a dry particle film on the glass surface.

The authors used pure O2 and N2 for the high-concentration range. For mixtures below 10% O2, air containing approximately 21% O2 replaced pure O2 to control dilute concentrations more accurately. The sample and gas lines were purged with nitrogen for at least 1 h before acquisition. The channel pressure drop was estimated at 0.5% of atmospheric pressure, so the authors calculated O2 partial pressure as 760 mmHg times the O2 fraction.

At zero field and room temperature, the authors applied approximately 0.5 W microwaves at the 2.87 GHz NV resonance. For continuous-wave spectra, ODMR contrast equaled the baseline fluorescence minus the dip minimum, divided by the baseline. Two-component Lorentzian fits supplied the contrast values.

For dynamic measurements, the LED was square-wave modulated at 20 Hz with an 85% duty cycle. Microwaves were modulated at 120 Hz with a 50% duty cycle. These conditions suit the particles’ dry-state T1 of about 1 ms, compared with 200 to 400 µs in water.

Computational demodulation separated the two amplitudes, and their MW/PL ratio supplied a real-time contrast value. Each point averaged eight 0.4 s traces, with data acquired every 4 s. The 1 h runs contained 900 points. The high optical duty cycle retained more excitation time than the roughly 50% duty cycle of typical light chopping. Ratioing the microwave and fluorescence amplitudes also suppressed common-mode drift.

Downward oxygen sweeps from 21% and 5% produced slight hysteresis. The 0 to 5% ascent had a raw slope of (-22.4 ± 2.6) × 10^-4 % mmHg^-1. The 5 to 1.1% descent had a shallower slope of (-12.4 ± 0.3) × 10^-4 % mmHg^-1. Repeated measurements at the same concentrations gave a repeatability of approximately 0.006% ODMR contrast. The authors state that incomplete removal of physisorbed O2 may cause the lower contrast on the descending sweep.

Equilibration also varied with concentration. A change from pure nitrogen to 10% O2 stabilized in about 2 min. Additions near 2.1% and 1.1% required about 10 min to stabilize. The continuous-wave response also appeared less linear below 20% O2, so the authors call for more precise concentration control to examine the transition from nitrogen to oxygen.

The authors found no prior demonstration of NV-based detection of molecular O2 in the gas phase. Earlier dissolved-O2 studies disagreed. A bulk diamond plate with shallow implanted NVs showed shortened relaxation, while nanodiamond particles showed no effect. This experiment instead measures gaseous O2 over dried particles.

The authors attribute the contrast decrease to less efficient NV polarization because paramagnetic oxygen collisions increase relaxation at the diamond surface. A measured increase in the NV T1 relaxation rate supports that explanation. Slow equilibration and oxygen-induced fluorescence quenching led them to suggest that physisorption contributes to the dynamics. They did not directly establish the adsorption mechanism.

The Stern-Volmer fluorescence plot was nonlinear over 0 to 760 mmHg. The authors note that this behavior is typical when fluorophores have accessible and inaccessible fractions, but saturation was not reached at 1 atm. They also consider a surface-driven shift in the NV⁻ to NV⁰ population. However, spectra under O2 and N2 showed no clear change in the 575 nm NV⁰ zero-phonon line relative to the 637 nm NV⁻ line. Only a slight NV⁻ luminescence decrease appeared. The gas environment, charge state, and adsorption chemistry therefore remain open questions.

Detection of oxygen from hydrogen peroxide

Section titled “Detection of oxygen from hydrogen peroxide”

The reaction test used 0.3% w/w, 88 mM H2O2 and 0.67 mg/mL catalase in a nitrogen-purged vessel. Nitrogen flowed at 4 mL/min toward the sensor. H2O2 additions of 176, 353, 529, and 882 µmol began at 300, 1000, 1700, and 2400 s. Each addition took 60 s.

Across the additions, the ODMR contrast decrease was linear with added H2O2. The response slope was (1.51 ± 0.13) × 10^-4% µmol^-1. Baseline noise was 0.0054%, giving a 3σ limit of 107 µmol H2O2. The lowest addition was therefore near, but above, the detection limit. Its 0.015% contrast reduction corresponded to an oxygen partial-pressure change of about 15 mmHg.

Catalase activity predicted that 176 µmol H2O2 would consume 88 µmol O2 in about 5.3 s. The corresponding undiluted pressure rise was about 50 mmHg, while nitrogen dilution reduced the measured change to about 15 mmHg. The minimum contrast occurred about 300 s after addition. The sensor therefore detected the oxygen burst, but its response was much slower than the underlying reaction.

Device implications and unresolved questions

Section titled “Device implications and unresolved questions”

The measured platform combines a dry 70 nm nanodiamond film, controlled gas flow, and double-modulated ODMR. Its present constraints are minute-scale equilibration, mild hysteresis, and sensitivity to surface gas chemistry. The authors also suggest that adsorbed oxygen could degrade the performance of single-NV diamond scanning probes.

A next sensor could use a narrower constriction to expose more particles to evolved O2. More uniform particle coverage could improve sampling across the optical field. The authors also propose testing smaller particles, including 40 nm diamonds, because they may improve sensitivity.

The reported result supports gas-phase O2 imaging at submicron scale as a possible use of NV sensors. The authors identify local oxygen gradients in catalysis, oxygen-related failure in MEMS, and coordination of nitrogen fixation with oxygenic photosynthesis in cyanobacteria as target applications. These uses remain proposals. The demonstrated instrument currently provides a 0 to 760 mmHg calibration and a projected limit near 8 mmHg, but it does not resolve reaction kinetics on the seconds timescale.

View Original Abstract

ABSTRACT Quantitative detection of paramagnetic molecular oxygen (O 2 ) in gas mixtures was achieved via optically detected magnetic resonance (ODMR) of negatively charged nitrogen‐vacancy (NV − ) centers in fluorescent nanodiamonds. Diamonds approximately 70 nm in diameter were deposited in a microfluidic channel, and the oxygen concentration was varied from 0% to 100% (0 to 760 mmHg O 2 partial pressure) by mixing O 2 and N 2 gases at ambient pressure. ODMR contrast was measured using a double‐modulation lock‐in detection scheme applied to both optical and microwave excitations. ODMR contrast decreases linearly with oxygen partial pressure, with a sensitivity coefficient k of (‐10.1 × ± 0.3) × 10 −4 % mmHg −1 . The oxygen detection limit of the experimental setup was estimated to be approximately 8 mmHg O 2 partial pressure (corresponding to ca. 1% O 2 ). Cycling O 2 content in the gas mixture in the range of 0‐5 % revealed a hysteresis and corresponding repeatability of 0.006 in percent ODMR contrast. Fluorescence quenching and slow response (several to tens of minutes) upon oxygen concentration changes suggest physisorption of gas molecules on the nanodiamond surfaces contributes to equilibration dynamics. Transient bursts of molecular oxygen generated by enzyme‐catalyzed decomposition of hydrogen peroxide were also detected.