Skip to content

Disordered Carbon Shells Enable Shape Dependent Fluorescence Enhancement in Nanodiamond Quantum Sensors Revealed by Correlative TEM–Fluorescence Microscopy

FieldValue
JournalAdvanced Functional Materials
Published2026-04-24
AuthorsParkarsh Kumar, Tom Vaillant, Mehran Kianinia, Haotian Wen, Adam Stewart
DOI10.1002/adfm.202531990

More than twofold volume-normalized brightness was measured in molten-KNO3-treated fluorescent nanodiamonds when the ηz aspect ratio exceeded six. For a 100 nm particle with a thickness of about 17 nm, corresponding to ηz = 6, the average NV center was about twice as bright as in an air-oxidized particle. Extrapolation to ηz = 8 and a thickness of about 12 nm indicated up to threefold brightness. The gain was concentrated in thin, flat particles, while bulkier particles showed similar brightness after either oxidation treatment.

The ensemble result was smaller because it averaged across shapes. KNO3-treated FNDs were approximately 20% brighter than air-oxidized FNDs under equal-mass ensemble photoluminescence measurements. That average does not describe the benefit available to a selected thin particle.

Both particle sets were nominally 100 nm, shared the same starting material, and underwent identical fragmentation milling and NV creation. The air-oxidized control was annealed in air at 450 °C for 1 h. The KNO3-treated sample was then treated with molten KNO3 at 500 °C for 1 h.

Previous work cited by the authors found minimal structural or morphological changes for FND thermal treatment between 450 and 500 °C. The comparison was therefore designed to isolate oxidation chemistry. Hot-water washing removed KNO3 and KNO2, which are highly soluble in water, with minimal residual salt reported.

Ensemble PL used equal dried-powder masses dispersed as 0.5 wt% suspensions in deionized water. Each sample was illuminated with a 532 nm laser through a 100x 1.3 NA oil-immersion objective. Emission passed through a 50 μm pinhole to a single-photon avalanche photodiode. Each map point averaged three 5 ms measurements, and no post-acquisition intensity normalization was applied.

HR-TEM showed that AO particles had little graphitic surface character, while KO particles carried an approximately 2 nm disordered, amorphous-like carbon shell. STEM-EELS at 300 kV gave a sharp carbon π* peak at 284.6 eV for AO surfaces. KO surfaces instead showed broad features from 284 to 288 eV, including C-H at 287.6 to 288.2 eV and C-O at 286.8 eV.

XPS C 1s analysis attributed 76% of the KO surface to defective sp3 carbon. Vacuum FT-IR found stronger C-O bands at 1100 to 1050 cm−1 and a C=O band at 1730 to 1680 cm−1 only in KO samples. These measurements establish a more defective and more oxidized surface, but they do not by themselves explain the brightness increase.

Single-particle correlation identifies the effect

Section titled “Single-particle correlation identifies the effect”

Ensemble spectroscopy could not connect brightness to individual particle geometry. TEMPL registered PL maps and TEM montages from the same area on a 10 nm Si3N4 membrane grid. Particle positions from TEM were aligned with PL peak centroids, and PL intensity was divided by particle volume to obtain VAPLI.

BF-TEM intensity was used with the Beer-Lambert law to reconstruct each particle’s three-dimensional thickness and shape. Shape was parameterized with size-invariant 3D Hu moments and converted into ηx, ηy, and ηz aspect ratios. An unsupervised machine-learning analysis related VAPLI to those descriptors.

The AO and KO size distributions were close. Mean diameters were 101 nm and 98 nm, respectively. The fractions below 50 nm were 11% for AO and 9% for KO, which argues against preferential loss of small particles.

Hu-moment ranges also overlapped, with H1 at 0.4 to 0.7 and H2 at 0.025 to 0.1. PC1 and PC2 explained 56.23% and 19.11% of shape variance, and most AO and KO particles clustered near zero. Only a few KO outliers departed from that shared distribution. The oxidation step therefore did not measurably make the KO population rounder or remove smaller particles.

Within each treatment, VAPLI increased with H1, H2, and ηz, confirming that flatter particles emitted more light per unit volume. The KO trend against ηz had a steeper slope than the AO trend. For equal flat shapes, KO particles produced more NV fluorescence per unit volume, whereas bulkier particles showed little separation between treatments.

MeasurementAO controlKO treatmentMeasurement condition
Mean particle diameter101 nm98 nmBF-TEM
Ensemble PL brightnessBaselineAbout 20% higherEqual mass, 0.5 wt%, 532 nm
Excited-state lifetime16 ns18 nsTri-exponential fit
ODMR contrastAbout 6%About 6%No external magnetic field
ODMR linewidthAbout 17 MHzAbout 17 MHzNo external magnetic field
T162 μs39 μsSpin measurement
T2*ComparableComparableDerived from Rabi measurements

Interference explains the shape dependence

Section titled “Interference explains the shape dependence”

The authors reject surface chemistry as the primary cause because shape-matched KO particles remained brighter. They argue that selective functional-group coverage on crystallographic facets would not produce this systematic size-independent shape dependence.

Instead, the authors propose constructive sub-wavelength interference between NV emission and reflection at the diamond-carbon interface. They assign the disordered shell a literature refractive index of 1.8 to 2.2, below that of diamond. The lower-index layer can induce constructive interference and increase the emitted fluorescence.

Lifetimes of 18 ns for KO and 16 ns for AO were fitted with a tri-exponential model. The two dominant components were 7 ns and 20 ns and were assigned to negatively charged and neutral NV centers. No new excited-state component appeared in KO, so the authors found no evidence for a new radiative pathway.

Earlier silica encapsulation was reported to raise luminescence by 1.9-fold and increase spin coherence from 1 to 3.5 μs to 52 to 87 μs. The current measured twofold gain at ηz = 6 is larger than that cited 1.9-fold value, although the studies use different particles and conditions. PMMA encapsulation, silica shells, and silica aerogel also showed increased emission in prior work cited by the authors.

The 1.8 to 2.2 refractive-index range comes from cited amorphous-carbon studies, not a direct optical measurement of these shells. Interference therefore remains the authors’ proposed mechanism, supported by particle-resolved correlations and prior low-index encapsulation results.

Spin readout remained largely stable. Both treatments produced about 6% ODMR contrast and about 17 MHz linewidths without an external magnetic field. T2* values were similar, which the authors interpret as evidence that intrinsic lattice impurities dominate decoherence in these FNDs.

T1 was shorter after KO treatment at 39 μs than after AO at 62 μs. The authors attribute the T1 difference to surface-chemistry-induced changes in near-surface NV centers. The treatment preserved ODMR contrast and T2*, while shortening T1.

Implications for sensor material selection

Section titled “Implications for sensor material selection”

For device design, particle geometry is a more useful selection variable than the oxidation label or ensemble brightness. A 100 nm FND near 17 nm thickness and ηz = 6 is the condition at which KO treatment yielded about twice the average NV brightness of AO. Bulkier FNDs had similar brightness regardless of oxidation, so the treatment offers no significant advantage there.

The process requires a 500 °C, 1 h molten KNO3 treatment and hot-water washing. Sensor development should track the 39 μs T1 value when replacing AO with KO, even though ODMR contrast and T2* remain comparable.

The threefold case at ηz = 8 is extrapolated, and the proposed interference mechanism uses a literature refractive-index range. Device claims should therefore distinguish the measured twofold gain from the extrapolated threefold limit. Broader qualification should also test morphology distributions because the optical response depends on particle thickness and aspect ratio.

View Original Abstract

ABSTRACT Fluorescent nanodiamonds (FNDs) containing nitrogen-vacancy (NV) centers are powerful quantum sensors capable of nanoscale optical and spin‐based detection of temperature and electric or magnetic fields. Improving their fluorescence brightness is critical for enhancing sensitivity, yet the underlying mechanisms remain poorly understood. Here, we show that treatment with strongly oxidizing molten produces an ultrathin (2 nm) disordered‐carbon shell on FND that significantly enhances NV emission. Using correlative transmission electron microscopy and photoluminescence (TEMPL), we directly link single‐particle morphology to optical performance and uncover previously inaccessible shape‐dependent behavior. For particles with aspect ratios exceeding six, volume‐normalized brightness increases by more than twofold compared with conventional air‐oxidized FNDs. Such enhancement does not arise from altered surface chemistry but from constructive sub‐wavelength interference at the diamond-carbon interface, enabled by the shell’s lower refractive index (1.8-2.2). Moreover, despite the presence of disordered carbon, spin properties, including ODMR contrast and times remain comparable to those of air‐oxidized FNDs. These results reveal the physical origin of brightness enhancement in ‐treated FNDs, establish low‐index interfacial shells as an effective strategy for enhancing NV emission, and provide a simple route to brighter FND quantum sensors.