Skip to content

Self‐Assembled Fluorescent Nanodiamond Layers for Quantum Imaging

FieldValue
JournalAdvanced Materials Interfaces
Published2026-02-05
AuthorsKatherine Chea, Erin S. Grant, Kevin J. Rietwyk, Hiroshi Abe, Takeshi Ohshima
DOI10.1002/admi.202500957
Citations1

Self-assembled fluorescent nanodiamond (FND) layers reached 16.8% physical coverage while producing NV photoluminescence above 10 kcps across 99.9% of the optimized sample. The layers resolved an effective magnetic field of 3.5 Gauss. Iron oxide deposition shortened a representative NV $T_1$ decay from 600 to 80 µs.

A positively charged polymer captured negatively charged FNDs through electrostatic attraction. Glass, silicon, or quartz was cleaned by sonication in acetone, ethanol, and deionized water, then exposed to UV ozone for 10 minutes. The Methods specify 5 minutes in 10 mg mL⁻¹ polyallylamine hydrochloride, followed by rinsing and nitrogen drying. The substrate was then immersed vertically in an aqueous FND suspension. Concentration ranged from 0.01 to 10 mg mL⁻¹, and immersion lasted from 1 to 1000 seconds. Silicon enabled reliable AFM analysis, while quartz allowed imaging through the substrate with low background PL. Quartz substrates were 0.5 mm thick and cut into 5 × 5 mm pieces. After assembly, each substrate was rinsed with water and dried on a 100 °C hotplate for 10 minutes. The FND source was 120 nm nominal high-pressure high-temperature nanodiamond. It received 2 MeV electron irradiation at a fluence of 1 × 10¹⁸ e⁻ cm⁻². It was annealed in argon at 900 °C for 2 hours, oxidized in air for 2.5 hours, and dispersed in water.

AFM separates single particles from aggregates

Section titled “AFM separates single particles from aggregates”

AFM used tapping mode with a Tap150AI tip and a typical 50 × 50 µm field. Scans contained 1024 × 1024 pixels at 0.25 Hz. Gwyddion and IgorPro analysis counted FNDs in 5 × 5 µm regions. Dynamic light scattering and AFM of isolated particles established a 120 nm hydrodynamic diameter. Disk-like FNDs usually attach on a larger face, so their AFM height falls below that diameter. A 55 nm z-height threshold separated singles from aggregates and stacks. Wide-field PL across 630 × 630 µm showed homogeneous red NV emission. A confocal image of about 9 × 9 µm resolved individual FNDs and small aggregates. The Supporting Information notes that the height rule counts neighboring, unstacked particles as singles.

Concentration and immersion time set surface density

Section titled “Concentration and immersion time set surface density”

At a fixed 100-second immersion, single and aggregate densities rose from 0.01 to 1 mg mL⁻¹. Both densities fell at 10 mg mL⁻¹, where aggregates became more common. The average surface-to-surface distance reached the 290 nm diffraction limit at 1 mg mL⁻¹. That reference applies to 520 nm green light and a 0.9 numerical aperture objective. At 0.1 mg mL⁻¹, the single-particle count increased with the square root of immersion time. Aggregate counts showed a clear increase only after 1000 seconds, about 16 minutes. The authors describe the saturation with a random sequential adsorption model, while noting that the model is too simple. They attribute the peak at 1 mg mL⁻¹ partly to aggregation in suspension. Sedimentation can affect FNDs larger than 100 nm over hours. Longer assembly may therefore raise total coverage while increasing aggregate formation.

The FND zeta potential was -38 mV at pH 6.4 and -12 mV at pH 4. The attached particle count rose from pH 6.4 toward pH 6.0 and 5.8, then fell to zero at pH 4. PAH remained protonated across this range because its amine pKa is above 8. The authors link the trend to protonation of oxygen-containing FND surface groups at lower pH. Adding 1 to 10 µM HCl slightly reduced particle repulsion and increased attachment in the tested range. Lower pH made the FND suspension less colloidally stable. Increasing NaCl from 10 µM to 10 mM changed the zeta potential from -43 mV to -31 mV. No systematic density change followed this added ionic strength. Attraction to PAH and repulsion between FNDs set coverage, because attached particles exclude nearby sites and lower repulsion raises aggregation risk.

Optimized layers collect PL across the surface

Section titled “Optimized layers collect PL across the surface”

The best tested condition used 1 mg mL⁻¹ FNDs and a 1000-second immersion without HCl. Adding HCl slightly reduced surface density at that concentration and immersion time. AFM measured 16.8% surface coverage, compared with 11.5% after 100 seconds at the same concentration. The authors state that 16.8% is well below dense monolayer coverage. Confocal PL exceeded 20 kcps on 91.4% of the substrate surface. A 10 kcps threshold produced 99.9% surface coverage, while the detector dark count was 1 kcps. The brightest FNDs produced more than 1 Mcps. This optical coverage does not mean that 99.9% of the surface was physically covered by diamond.

The authors drop-cast Fe₂O₃ micro- and nanoparticles onto FND-coated quartz. A custom wide-field quantum diamond microscope measured ODMR spectra and $T_1$ decay traces before and after deposition. Each analyzed region was about 8 × 8 µm, and the histogram used 625 pixels over 8.1 × 8.1 µm. Static fields split the NV spin sublevels by the Zeeman effect, broadening ODMR for randomly oriented FND axes. Before iron oxide deposition, the resonance sat just below 2.87 GHz with a 23 MHz full width at half maximum. After deposition, the width increased to 36 MHz and the center shifted to 2.86 GHz. The authors suggest that light-induced heating of the iron oxide caused the center-frequency shift. The estimated effective field peaked at 0.5 Gauss without particles and 1.0 Gauss with particles. The maximum was 3.5 Gauss beside the largest aggregate, approaching zero at distances of 10 to 20 µm. These values agree with prior measurements for similar particles, but the effective field remains a relative estimate.

Fluctuating fields resonant with NV spin transitions shorten $T_1$, allowing a separate map of magnetic noise. A representative decay changed from 600 µs without particles to 80 µs after deposition. A single-exponential fit gave those values, and the histogram peak fell from 600 µs to about 60 µs. The strongest decreases occurred near smaller particles scattered across the upper half of the field. Grey map regions could not be fitted, probably because local noise pushed $T_1$ below the 10 µs measurement resolution. The authors associate this relaxation with fluctuating magnetic moments in the nanoparticles. They distinguish that noise from the static field near the larger aggregate. Particle size can place the iron oxide in superparamagnetic or weakly ferromagnetic behavior, according to the authors.

Implications for sensor material selection

Section titled “Implications for sensor material selection”

The coated layer reached 16.8% physical coverage rather than a dense monolayer. For static-field work, ODMR broadening provides a relative field estimate rather than a full vector measurement. For noise work, some pixels fell below the 10 µs $T_1$ measurement resolution. Quartz suits transmission imaging and low background PL, while silicon supported surface analysis. The authors propose broader use on rigid and flexible surfaces, but their magnetic demonstration used quartz. Bulk diamond imaging has used millimeter-scale chips that are costly and difficult to integrate with other materials. Earlier FND work produced sparse arrays, including a covalent patterned-silicon array with a yield above 90%, and inhomogeneous drop-cast layers. The self-assembled layer instead concentrates sensing material close to the surface target and limits signal from particles farther along the optical path. The authors expect dense PL coverage to raise the NV signal per unit area, although physical coverage remains partial. For device design, suspension concentration, immersion time, surface chemistry, and colloid stability must be controlled together. The results support FND coatings when surface integration and large-area coverage matter more than imaging with individual NV centers.

View Original Abstract

ABSTRACT The nitrogen‐vacancy (NV) center in diamond is emerging as a powerful tool for imaging magnetic and electric signals at the microscale and below. However, most imaging demonstrations thus far have relied on costly, millimeter‐sized bulk diamond substrates, which cannot be easily scaled or integrated with other materials. Here, we report a scalable method for fabricating NV‐containing dense and homogenous fluorescent nanodiamond (FND) layers through electrostatic self‐assembly and demonstrate the utility of the FND layers for magnetic imaging. We investigate the effect of FND concentration in suspension, substrate immersion time, and solvent pH on the FND density on the substrate. We identify optimized self‐assembly conditions that maximize the FND density while minimizing aggregation. Using FND layers on a quartz substrate, we demonstrate magnetic field and magnetic noise imaging at the microscale, based on NV optically detected magnetic resonance magnetometry and T 1 relaxometry, respectively. Our results provide a direction for the development of cost‐effective and scalable FND layers and surface coatings. This paves the way for on‐demand quantum sensing and imaging on a broad range of surfaces based on NV centers and other diamond quantum emitters.