Skip to content

The properties of the nitrogen-vacancy center in milled chemical vapor deposition nanodiamonds

FieldValue
JournalMaterials for Quantum Technology
Published2026-02-12
AuthorsAlessandro Mameli, Giannis Thalassinos, Marco Capelli, Johannes Ackermann, Edwin Harrop Mayes
DOI10.1088/2633-4356/ae4565
Citations2

Ball milling produced fluorescent nanodiamonds (FNDs) with an average spin relaxation time T₁ of 4.7 ± 1.6 ms. The parent bulk diamond gave 3.2 ± 0.7 ms. Commercial high-pressure high-temperature (HPHT) nanodiamonds gave 0.17 ± 0.01 ms. Prior work cited in the paper reports bulk T₁ values of 2 to 5 ms and typical HPHT FND values on the order of 100 μs. These measurements show that milling preserved bulk-like T₁ relaxation in the tested CVD particles.

CVD parent material and controlled nitrogen

Section titled “CVD parent material and controlled nitrogen”

Three single-crystal CVD samples were grown on 3 × 3 mm² HPHT substrates polished to Ra below 0.5 nm. The substrates were approximately 200 μm thick. Each received 10 min H₂/O₂ and 10 min H₂ plasma treatments at 800 °C. Growth used 2.67% methane in hydrogen at 210 mbar and 800 to 830 °C for 45 h. The final CVD plate was approximately 500 μm thick. A 2.45 GHz, 6 kW microwave generator powered the reactor. The plasma nitrogen to carbon ratio was 6500 ppm.

Electron irradiation at 2 MeV and 1 × 10¹⁸ e⁻ cm⁻² supplied vacancies. Annealing at 900 °C for 2 h in vacuum created NV centers. UV-visible absorption indicated 1.81 to 2.05 ppm substitutional nitrogen and 0.25 to 0.28 ppm negatively charged NV centers. One sample remained bulk, and two were milled.

SampleSubstitutional Nₛ⁰Electron dose (e⁻ cm⁻²)NV⁻ concentration
Bulk CVD2.05 ppm1 × 10¹⁸0.27 ppm
CVD FND1.85 ppm1 × 10¹⁸0.26 ppm
HPHT FND100 to 200 ppm1 × 10¹⁹1 to 3 ppm

The HPHT defect concentrations came from the manufacturer, and its irradiation dose was about ten times higher. The comparison is not milling-only, because nitrogen concentration and irradiation dose also differ. Commercial HPHT FNDs of similar size served as the reference.

Two bulk samples were milled with 15 stainless-steel balls, 4 mm in diameter, in 5 ml isopropanol. Milling ran for 6 h at 50 Hz in a vibrating ball mill. One bulk sample was retained as a control. Steel was chosen because oxidizing acids can remove metal debris and surface nondiamond carbon. Purification used 100 ml of 96% sulfuric acid followed by 50 ml of 65% nitric acid. The mixture was stirred at 140 °C for 12 h, then washed by repeated centrifugation until neutral pH. Centrifugation ran at 15,000 rpm for 1 h per cycle. TEM showed particles spanning several hundred nanometers to below 100 nm. DLS detected sizes from below 100 nm to above 1 μm, with most volume-weighted material near 100 nm.

Confocal photoluminescence (PL) spectra used 520 nm excitation, while paired 405 nm and 520 nm spectra separated NV⁻ from neutral NV. Charge ratios came from least-squares fitting of normalized spectra, with about 30 particles examined per FND type. The NV⁻ ratio is NV⁻ photoluminescence divided by total NV photoluminescence. The bulk CVD ratio was 60%. CVD FNDs averaged 31%, with individual values from 1% to 55%. HPHT FNDs averaged 46%, with values from 25% to 64%. After milling, surface acceptors can trap donor electrons, which the authors propose as the cause of the lower ratio. Their estimate gives about 1,000 surface acceptors and 100 donor electrons per 100 nm CVD particle. Concentration estimates gave about 100 donor electrons per NV in HPHT material, compared with about 7 in the CVD samples. This makes the ratio sensitive to particle size and surface chemistry.

Time-resolved PL traces were fitted with a double exponential. The amplitude-weighted average PL lifetime was 11 ns for bulk CVD, 29 ns for CVD FNDs, and 17 ns for HPHT FNDs. Both nanodiamond samples showed multi-exponential decay. CVD FNDs had a 34 ns slow tail, compared with 28 ns for HPHT FNDs. A 0.6 ns component within the first 1.5 ns was absent from bulk. The authors associated it with NVs very close to particle surfaces. Charge ratio and lifetime did not correlate in CVD particles. The authors attributed the longer CVD lifetime to lower effective refractive index and a higher NV⁰ contribution. Lower nitrogen and vacancy-defect concentrations were also identified as likely contributors.

Pulsed 514 nm excitation and microwave control were used for optically detected magnetic resonance (ODMR). The authors fitted each spectrum with two Lorentzians. Average contrast was about 3% for bulk CVD, CVD FNDs, and HPHT FNDs. Only 38% of CVD FNDs produced measurable ODMR, compared with 73% of HPHT FNDs. The authors relate this low yield to the lower NV⁻ content in CVD particles.

The zero-field splitting D departed from the typical 2870 MHz bulk value by as much as 2.5 MHz. This corresponded to about 200 MPa strain. The cited laser-heating shift is 74 kHz K⁻¹, but it would affect all particles similarly. The authors therefore identify milling strain as the likely dominant cause. The transverse parameter 2E averaged 7.7 ± 1.9 MHz for CVD FNDs and 11.2 ± 1.6 MHz for HPHT FNDs. Bulk CVD gave 5.1 ± 0.1 MHz. The CVD value was 31% lower than HPHT. The authors attribute the difference to fewer charged defects in CVD material, while considering milling strain as a separate source of increased 2E. Bulk birefringence was fairly homogeneous at 4 × 10⁻⁵ to 9 × 10⁻⁵. The authors therefore associate the larger post-milling spread with milling-induced strain. Power broadening made the measured linewidths larger than the intrinsic linewidths. A direct T₂* improvement was not demonstrated for the lower 2E condition.

Spin relaxation in ensembles and single particles

Section titled “Spin relaxation in ensembles and single particles”

T₁ traces used 514 nm excitation, with signal traces corrected by subtracting microwave-on reference traces. FND ensemble traces came from dense regions containing more than 10 particles. Single-exponential fits gave 4.7 ± 1.6 ms for CVD FNDs and 3.2 ± 0.7 ms for bulk CVD. HPHT FNDs gave 0.17 ± 0.01 ms. The shorter bulk value may reflect incomplete initialization of spins across the collection volume. Most CVD material by volume was near 100 nm, but larger particles could dominate the ensemble decay. One particle estimated below 250 nm gave a T₁ of 3.7 ± 0.9 ms. This measurement confirms millisecond relaxation in at least one particle within the diffraction-limited focal spot. Within the tested samples, milling therefore had little effect on ensemble T₁.

HPHT FNDs were estimated to be about 10 times brighter. Their NV concentration was 1 to 3 ppm, compared with 0.26 ppm for CVD FNDs. The HPHT sample also had a higher NV⁻ ratio and shorter T₁. HPHT particles are available in kilogram quantities, whereas CVD particles are not currently available at that scale. The authors consider large-scale CVD production feasible as prices decline, but they expect costs to remain above HPHT. The authors identify surface chemistry and defect engineering as the route to stable NV⁻ without compromising coherence. They also propose raising nitrogen or adding nonparamagnetic electron donors to improve charge state and brightness. However, nitrogen above 100 ppm can shorten transverse spin relaxation through the nitrogen spin bath. For T₁ relaxometry, sensitivity scales with the relaxation rate, so milled CVD nanodiamonds are candidates for applications prioritizing T₁ over brightness. For ODMR magnetometry or thermometry, the tested CVD particles remain limited by low estimated brightness and poor ODMR signal yield. Material selection therefore depends on whether the sensor prioritizes relaxation time, charge stability, or photon count rate.

The paper needs nitrogen-containing single-crystal CVD parent material for fluorescent nanodiamonds used in quantum sensing. The diamond materials catalog lists single-crystal MPCVD material from 0.1 to 500 μm thick, including the paper’s approximately 500 μm parent thickness. The nitrogen content, irradiation, annealing, and milling requirements need separate specification through engineering support.

View Original Abstract

Abstract Fluorescent nanodiamonds (FNDs) containing negatively charged nitrogen-vacancy (NV − ) centers are vital for many emerging quantum sensing applications from magnetometry to intracellular sensing in biology. However, developing a scalable fabrication method for FNDs hosting color centers with consistent bulk-like photoluminescence (PL) and spin coherence properties remains a highly desired but unrealized goal. Here, we investigate optimized ball milling of single-crystal diamonds produced via chemical vapor deposition (CVD) and containing 2 ppm of substitutional nitrogen and 0.3 ppm of NV − to achieve this goal. The NV charge state, PL lifetime, and spin properties of bulk CVD diamond samples are directly compared to milled CVD FNDs and commercial high-pressure high-temperature (HPHT) FNDs. We find that on average, the relative contribution of the NV − charge state to the total NV PL is lower and the NV PL lifetime is longer in CVD FNDs compared to HPHT FNDs, both likely due to the lower N s 0 concentration in CVD FNDs. The CVD bulk and CVD FNDs on average show similar average T 1 spin relaxation times of 3.2 ± 0.7 ms and 4.7 ± 1.6 ms, respectively, compared to 0.17 ± 0.01 ms for commercial HPHT FNDs. Our results demonstrate that ball milling of CVD diamonds enables the large-scale fabrication of NV ensembles in FNDs with bulk-like T 1 spin relaxation properties.

  1. 2021 - Quantum networks based on color centers in diamond [Crossref]
  2. 2020 - Sensitivity optimization for NV-diamond magnetometry [Crossref]
  3. 2018 - Probing condensed matter physics with magnetometry based on nitrogen-vacancy centres in diamond [Crossref]
  4. 2021 - Toward quantitative bio-sensing with nitrogen-vacancy center in diamond [Crossref]
  5. 2024 - Quantum sensing and light guiding with fluorescent nanodiamond‐doped PVA fibers [Crossref]
  6. 2018 - Microscale-resolution thermal mapping using a flexible platform of patterned quantum sensors [Crossref]
  7. 2020 - Electrospun nanodiamond-silk fibroin membranes: a multifunctional platform for biosensing and wound-healing applications [Crossref]
  8. 2019 - Quantum sensing in a physiological‐like cell niche using fluorescent nanodiamonds embedded in electrospun polymer nanofibers [Crossref]
  9. 2023 - Hybrid quantum nanophotonic devices with color centers in nanodiamonds [Crossref]
  10. 2019 - Quantum nanophotonics with group IV defects in diamond [Crossref]