Solution-phase fluorination of nanodiamond: near-surface NV$^-$ activation and spin relaxation

AI summary

Overview

Research area: Quantum materials chemistry — surface functionalization of fluorescent nanodiamonds (FNDs) hosting nitrogen–vacancy (NV) centres, and its effect on charge-state stability and spin relaxation.

Technical level: Intermediate. The paper combines wet-chemistry synthesis with standard materials characterization (FTIR, XPS, EDS, Raman, PL) and optically detected magnetic resonance (ODMR) relaxometry; readers need some familiarity with NV physics and surface spectroscopy.

Scope (one sentence): The paper introduces and compares two solution-phase fluorination routes for 140 nm HPHT nanodiamonds and measures how the resulting C–F surface chemistry affects the NV⁻ charge fraction and the spin–lattice relaxation time T₁.

What This Paper Is About

Shallow NV centres in nanodiamonds are excellent nanoscale sensors, but the same surface proximity that makes them sensitive also exposes them to spin noise and to charge-state instabilities that convert NV⁻ into NV⁰. Fluorine termination is theoretically favoured because its positive electron affinity stabilizes NV⁻, yet existing fluorination methods rely on aggressive reagents (F₂, ClF₃, plasmas) or give only moderate gains in nanodiamond. The authors develop two mild, solution-phase fluorination chemistries and ask whether high fluorine coverage actually delivers both better charge stability and longer spin lifetimes.

Key Contributions

1. Two wet-chemical fluorination routes demonstrated on nanodiamond. A xenon difluoride (XeF₂) route forming C–F bonds by decarboxylation through a radical mechanism, and a Balz–Schiemann route that converts surface amino groups (previously installed by Hofmann degradation) into fluorine via diazotisation.
2. Multi-technique verification of covalent C–F bonding and lattice preservation. FTIR, XPS, EDS, Raman and PL were used to confirm fluorination, quantify fluorine loading, and show that the diamond lattice survives the treatment.
3. A near-unity NV⁻ fraction. Both routes gave up to ∼90% NV⁻ on average and approaching 100% in fluorine-rich regions, which the authors describe as among the highest reported for surface-terminated nanodiamonds of this size, and specifically for fluorine termination.
4. A direct measurement showing that charge stability and spin lifetime do not improve together. Frequency-domain relaxometry yielded T₁ = 733 ± 56 µs (XeF₂) and 712 ± 20 µs (Balz–Schiemann), versus 1173 ± 123 µs for the as-received material.

Main Findings

• Different fluorine loadings for the two routes. XPS measured 2.6 at.% F after the XeF₂ treatment and 0.7 at.% F after the Balz–Schiemann reaction. Assuming spherical particles and an XPS information depth of 5–10 nm, this corresponds to roughly 20%–40% of a monolayer (XeF₂) and 5%–10% (Balz–Schiemann). The lower Balz–Schiemann yield is attributed partly to incomplete Hofmann degradation (unreacted COOH groups visible in FTIR) and partly to diazonium salt stability, which is influenced by the counterion.

• Covalent C–F bonds identified spectroscopically. FTIR showed new bands at 1255 cm⁻¹ (ND_NH₂-F) and 1265 cm⁻¹ (ND_XeF₂), a feature near 800 cm⁻¹ in both, and increased intensity around 1100 cm⁻¹, consistent with C–F stretching stiffened by the diamond lattice. XPS added a C 1s component at 288.2 eV and F 1s components at 688.2 eV and 686 eV, assigned to covalent and semi-ionic C–F environments. Oxygen- and hydrogen-related bands decreased, indicating conversion of pre-existing surface groups.

• The diamond lattice is not damaged. Raman showed the diamond peak at 1335.5 cm⁻¹ in all samples with the G band at 1613 cm⁻¹. Full widths at half maximum of the diamond band were 10.4 cm⁻¹ (COOH), 9.6 cm⁻¹ (XeF₂) and 10.8 cm⁻¹ (Balz–Schiemann), confirming the reactions were mild.

• Fluorination increases sp² carbon, but differently in each route. The sp³:sp² peak-area ratio fell from ≈20 in the as-received powder to ≈4 for ND_XeF₂ and ≈0.1 for ND_NH₂-F. The strong Balz–Schiemann sp² signal is partly assigned to a by-product that survived repeated water washing, and partly to an artefact of resonant excitation of adsorbed aromatic species such as pyridine: the G-band enhancement was much weaker under 785 nm excitation than under 325 nm.

• High NV⁻ fraction. PL spectra recorded under 532 nm excitation across two orders of magnitude in power density were decomposed into NV⁻ and NV⁰ components plus background terms; the resulting NV⁻ fraction reached up to ∼90% on average and approached 100% in fluorine-rich regions. (The supplied text is truncated within this PL analysis, so the full power-dependent dataset is not available in the excerpt.)

• Fluorine distribution is laterally inhomogeneous. EDS at 8 kV gave an average F content of 4.6% in the homogeneously covered edge region and 4.7% in the interior, where fluorine appears in patches within a more hydrophobic environment. Element maps show that where fluorine is higher, oxygen is lower, confirming that fluorination replaces oxygen-containing groups; fluorine-rich particles also aggregate more readily.

• Spin relaxation shortens relative to the starting material. Zero-power-extrapolated T₁ values from frequency-domain relaxometry were 733 ± 56 µs (XeF₂), 712 ± 20 µs (Balz–Schiemann) and 1173 ± 123 µs (as-received). The fluorinated values are described as several times those reported for commercial HPHT nanodiamonds, yet shorter than the untreated powder.

• The central trade-off. Fluorination activates near-surface NV⁻ centres — the ones most exposed to surface noise, but also the ones that dominate relaxometric sensing — so improved charge-state stability comes at the cost of a shorter spin lifetime.

Methodology in Plain English

Starting material. Commercial HPHT diamond particles, 140 nm in diameter with 3 ppm NV content (Adámas Nanotechnologies).

Route 1 — Balz–Schiemann. Amino groups were first installed on the surface using a Hofmann degradation sequence (thionyl chloride, then ammonia in dioxane, then NaOH and bromine). The amino-terminated particles were then diazotised with sodium nitrite in the presence of HF/pyridine at 0–2 °C and heated to 87–88 °C, converting surface NH₂ into C–F. The product is labelled ND_NH₂-F.

Route 2 — XeF₂. Dried nanodiamond was sealed with 100 mg XeF₂ in deuterated chloroform with HF/pyridine and heated at 100 °C for 70 hours. Deuterated solvent was chosen to suppress side reactions that would form C–H bonds instead of C–F. The product is labelled ND_XeF₂. In both routes HF/pyridine acts as catalyst and solvent rather than as the fluorine source, since hydrated fluoride is a poor nucleophile and anhydrous HF alone does not fluorinate diamond effectively.

Checking the chemistry. FTIR identified C–F vibrational bands; XPS quantified surface composition and bonding environments; EDS mapped lateral fluorine distribution; Raman tested lattice integrity at two excitation wavelengths (325 nm and 785 nm) to separate true disorder from resonance artefacts.

Measuring the quantum properties. Photoluminescence under 532 nm excitation was fitted as a sum of NV⁻ and NV⁰ spectral components plus background, giving the charge-state ratio. ODMR was performed on dried samples using a 520 nm laser (220.5 mW, about 0.78 W cm⁻²) and microwaves near 2870 ± 70 MHz. Rather than pulsed sequences, T₁ was obtained by frequency-domain relaxometry: the microwave amplitude was modulated and the lock-in response recorded while the modulation frequency was swept from about 10 Hz to 20 kHz. The response peaks at a characteristic frequency f_max; plotting f_max against laser power (four powers from 142 mW to 354 mW) and extrapolating to zero power gives the intrinsic T₁ = 1/(2πf₀), removing the contribution of continuous illumination.

Why This Matters

Fluorine termination is one of the most theoretically favoured surface chemistries for keeping shallow NV centres negatively charged, but practical routes to it have been harsh, poorly scalable, or only moderately effective on nanodiamonds. This work shows that ordinary benchtop solution chemistry — no F₂ gas, no ClF₃, no plasma — can push the NV⁻ fraction toward 90% and near 100% locally. It also delivers an important caution: the highest charge stability measured here coincided with a shorter T₁ than the untreated powder, so surface engineers cannot assume that one figure of merit tracks the other.

Real-world applications:

• Nanoscale magnetometry with shallow NV centres, including single-centre detection at the nT scale and nm-resolution magnetic imaging.
• Nanodiamond thermometry, where sensitivities of ∼130 mK Hz⁻¹/² have been reported in nanodiamonds.
• All-optical T₁ relaxometry for detecting paramagnetic free radicals, including in single living cells and embryos and for pH sensing over the range 3–7 with responsive coatings.
• Charge-state-based sensing of local electrochemical potential, with nanodiamond NV ensembles resolving swings as small as ∼20 mV.

Industry relevance: The routes use commercially available reagents and standard laboratory glassware/Teflon ware, which matters for scaling fluorescent nanodiamond production and for manufacturers who cannot operate fluorine gas or plasma lines. The finding that fluorine-rich particles aggregate also has direct implications for colloidal formulation and dispersion of FND products.

Future Directions

• Fix the Balz–Schiemann by-product. The sp²-rich residue that survived repeated washing (sp³:sp² ≈ 0.1) needs a purification or reaction redesign, and counterion choice for the diazonium salt is an identified lever for improving yield.
• Decouple coverage from relaxation. Since fluorination raised NV⁻ fraction while lowering T₁ below the as-received 1173 ± 123 µs, the next step is a termination that stabilizes charge without introducing the spin bath responsible for the shortening.
• Separate surface from bulk contributions. Determining whether the measured T₁ reflects predominantly near-surface centres — the ones that dominate ensemble relaxometry — would clarify how much of the lifetime penalty is intrinsic to fluorine termination.
• Combine with complementary chemistries. Mixed N/O terminations and inorganic shells (silica, Al₂O₃) have independently improved NV stability and pushed T₁ to 4.7 ms in 100 nm nanodiamonds, ∼2 ms after molten KNO₃ oxidation and 1 ms in 50–60 nm particles with silica coating; whether fluorination stacks beneficially with these approaches is untested here.

Target Audience

Researchers in quantum sensing and quantum materials who work with NV centres in nanodiamond; surface and materials chemists interested in mild routes to diamond functionalization; groups developing fluorescent nanodiamonds for bioimaging, relaxometry-based radical detection, or nanoscale thermometry and magnetometry; and graduate students seeking a worked example of linking surface spectroscopy to spin-physics measurements. Readers focused on high-pressure synthesis or on bulk diamond electronics will find less of direct relevance.

Authors’ abstract

Nanodiamonds hosting luminescent point defects, known as fluorescent nanodiamonds (FND), are a leading platform for quantum technology. The nitrogen-vacancy (NV) centre is the most intensively studied of these; its negatively charged state (NV$^-$) can serve as a qubit at room temperature, and its stability is governed by surface functional groups. We present two solution-phase fluorination routes for stabilising NV$^-$: direct C-F bond formation by decarboxylation with xenon difluoride via a radical mechanism, and the Balz-Schiemann reaction, which replaces surface amino groups with fluorine. The two routes were compared by infrared, X-ray photoelectron, energy-dispersive X-ray, Raman and photoluminescence spectroscopy. Both gave a high NV$^-$ fraction, up to $\sim$90% on average and approaching 100% in fluorine-rich regions, which to our knowledge is among the highest reported for surface-terminated nanodiamonds of this size and, in particular, for fluorine termination. Frequency-domain relaxometry shows that the fluorinated particles retain a long spin-lattice relaxation time, $733\pm56$ and $712\pm20$ $μ$s for the XeF$_2$ and Balz-Schiemann routes, several times the values reported for commercial HPHT nanodiamonds, although shorter than the $1173\pm123$ $μ$s of the as-received material. Charge-state stability and spin lifetime therefore do not improve together: fluorination activates near-surface NV$^-$ centres, which are the most exposed to surface noise but also the ones that dominate relaxometric sensing.